A ship all-weather multi-scene acquisition method and device

Abstract

The application provides a kind of ship all-weather multi-scene acquisition method and device, it is related to the technical field of data acquisition, method includes: obtaining offshore environment data, ship attitude data and the historical trajectory data of preset target;Wind wave data is generated based on offshore environment data, and wind wave data is used to represent the influence of wind wave on the movement of ship swing and preset target;The historical trajectory data is processed to form the movement trend trajectory of preset target;Combining wind wave data and movement trend, the simulation position of preset target is corrected, and the pre-track of preset target is generated;The pre-vector corresponding to the acquisition equipment is converted from pre-track, and the attitude compensation is carried out in combination with ship attitude data to form the turntable control quantity of acquisition equipment;The acquisition equipment is controlled based on turntable control quantity to collect preset target.The application can improve the collection efficiency for high-speed small target.

Description

Technical Field

[0001] This invention relates to the technical field of data acquisition, specifically to a method and apparatus for data acquisition in various scenarios and around the clock on ships. Background Technology

[0002] Current shipboard maritime target acquisition typically relies on shipborne electro-optical gimbals, visible light cameras, low-light cameras, infrared cameras, and inertial image stabilization structures to achieve long-distance observation, capturing images of ships, floating objects, and low-flying targets in daylight, nighttime, rain, fog, and low-light conditions. To reduce image jitter caused by changes in hull roll, pitch, and heading, some shipborne acquisition equipment uses a turntable mechanism to drive the optical payload for azimuth and pitch adjustments, and combines this with components such as gyroscopes, encoders, and motors to achieve a certain degree of image stabilization. These methods can meet the continuous observation needs of conventional navigation targets, low-speed targets, or relatively stable sea conditions, and can also adapt to changes in lighting and visibility through different camera modes.

[0003] However, in scenarios involving the acquisition of high-speed, small targets such as unmanned surface vessels, low-altitude small aircraft, and rapidly drifting targets on the sea surface, the targets are small in size, change speed rapidly, and are easily affected by wind speed, wind direction, wave height, wave direction, and wave period, resulting in short-term deviations. Simultaneously, the vessel itself experiences roll, pitch, and bow disturbances in high winds and waves. Conventional turntables often rely on current image deviations or current attitude feedback for follow-up compensation, failing to combine marine environmental data with historical target trajectory data to predict the target's future position. Thus, when switching between visible light, low-light, and infrared cameras, or when the target passes through rain, fog, waves, or the edge of the field of view, the turntable's pointing can easily lag behind the target's actual movement. This forces the acquisition equipment to re-search for the target, leading to problems such as short-term target loss, discontinuous acquisition across scenarios, and a reduced effective acquisition rate for high-speed, small targets. Summary of the Invention

[0004] This invention provides a method and apparatus for all-weather, multi-scenario data acquisition from ships, which can improve the acquisition efficiency for high-speed, small targets.

[0005] In a first aspect, the present invention provides a method for collecting data from multiple scenarios in all weather conditions on a ship, the method comprising: Acquire marine environmental data, ship attitude data, and historical trajectory data of preset targets; Wind and wave data are generated based on the marine environment data, and the wind and wave data is used to characterize the impact of wind and waves on the hull roll and the motion of the preset target. The historical trajectory data is processed to form the motion trend trajectory of the preset target; By combining the wind and wave data with the motion trend track, the simulated position of the preset target is corrected, and the pre-trajectory of the preset target is generated; The pre-trajectory is converted into a pre-pointing quantity corresponding to the acquisition device, and attitude compensation is performed in combination with the ship attitude data to form the turntable control quantity of the acquisition device. The acquisition device is controlled to acquire data from the preset target based on the turntable control quantity.

[0006] In a second aspect of the invention, a ship all-weather multi-scene data acquisition device is provided. The device is used to execute a ship all-weather multi-scene data acquisition method as described in any of the above embodiments. The device includes an acquisition module, a processing module, and an output module, wherein: The acquisition module is used to acquire marine environmental data, ship attitude data, and historical trajectory data of preset targets; The processing module is used to generate wind and wave data based on the marine environment data, and the wind and wave data is used to characterize the impact of wind and waves on the hull roll and the motion of the preset target; The processing module is used to process the historical trajectory data to form the motion trend trajectory of the preset target; The processing module is used to combine the wind and wave data with the motion trend track to correct the simulated position of the preset target and generate the pre-trajectory of the preset target; The processing module is used to convert the pre-trajectory into a pre-pointing quantity corresponding to the acquisition device, and combine it with the ship attitude data to perform attitude compensation, thereby forming the turntable control quantity of the acquisition device. The output module is used to control the acquisition device to acquire data from the preset target based on the turntable control quantity.

[0007] In a third aspect of the invention, an electronic device is provided, including a processor and a memory, the memory storing a computer program, wherein the computer program, when executed by the processor, implements a ship all-weather multi-scene data acquisition method as described in any of the preceding claims.

[0008] In a fourth aspect, the present invention provides a non-transitory computer-readable storage medium storing instructions that, when executed, perform a ship all-weather multi-scenario data acquisition method as described in any of the preceding claims.

[0009] In summary, one or more technical solutions provided in the embodiments of the present invention have at least the following technical effects or advantages: This invention improves the efficiency of acquiring data from high-speed, small targets primarily by moving beyond the passive following method of conventional acquisition equipment that relies on current image deviations. Instead, it integrates marine environmental data, ship attitude data, and historical trajectory data of the target into a single processing chain. Marine environmental data is used to generate wave data reflecting wind and wave disturbances. Historical trajectory data is used to form the target's motion trend trajectory. The wave data is then used to correct the simulated position, resulting in a pre-trajectory of the target. This pre-trajectory is then converted into a pre-pointing variable executable by the acquisition equipment. Combined with changes in ship roll, pitch, and heading, this forms a turntable control variable, enabling the acquisition equipment to perform azimuth, pitch, and image stabilization compensation before the target actually deviates from the field of view. In scenarios with small targets and rapidly changing speeds, such as unmanned surface vessels, low-altitude small aircraft, and fast-drifting targets on the sea surface, the acquisition equipment reduces the number of times it needs to re-search for the target. It maintains continuous pointing and acquisition under conditions of switching between visible light, low-light, and infrared cameras, as well as interference from rain, fog, waves, and field-of-view edges, thereby improving the effective acquisition rate of high-speed, small targets and cross-scenario acquisition efficiency. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating a method for collecting data from multiple scenarios in all weather conditions on a ship, as disclosed in an embodiment of the present invention. Figure 2 This is a schematic diagram of a data acquisition device disclosed in an embodiment of the present invention; Figure 3 This is a schematic diagram of a module of a ship-based all-weather, multi-scenario data acquisition device disclosed in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention.

[0011] Explanation of reference numerals in the attached drawings: 301, acquisition module; 302, processing module; 303, output module; 401, processor; 402, communication bus; 403, user interface; 404, network interface; 405, memory. Detailed Implementation

[0012] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0013] In the description of the embodiments of the present invention, words such as "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "for example" or "for instance" in the embodiments of the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Rather, the use of words such as "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0014] In the description of the embodiments of the present invention, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0015] Current shipboard maritime target acquisition relies heavily on shipboard photoelectric turntables and acquisition channels such as visible light cameras, low-light cameras, and infrared cameras. Conventional target observation is achieved through azimuth rotation, pitch adjustment, and inertial image stabilization. However, in scenarios involving high-speed, small targets such as unmanned surface vessels, low-altitude small aircraft, and rapidly drifting targets on the sea surface, the targets are small in size and move rapidly. They are also affected by wind speed, wind direction, wave height, wave direction, wave period, as well as the ship's roll, pitch, and bow disturbances. Conventional turntables usually follow and compensate based on the current image deviation or current attitude, lacking joint prediction of marine environmental data and historical trajectory data. This leads to problems such as pointing lag, short-term target loss, and discontinuous acquisition across scenarios when the acquisition equipment is obscured by rain and fog, interfered with by waves, is at the edge of the field of view, or during acquisition channel switching.

[0016] The present invention discloses a method for collecting data on ships in all weather conditions and multiple scenarios, which is applied to a server. The server includes, but is not limited to, electronic devices such as mobile phones, tablets, wearable devices, and PCs (Personal Computers), and can also be a backend server running the method for collecting data on ships in all weather conditions and multiple scenarios. The server can be implemented using a standalone server or a server cluster composed of multiple servers.

[0017] This embodiment discloses a method for collecting data from ships in all weather conditions and multiple scenarios, referring to... Figure 1 It includes the following steps: S110 acquires marine environmental data, ship attitude data, and historical trajectory data of preset targets.

[0018] S120 generates wind and wave data based on marine environmental data.

[0019] S130 processes historical trajectory data to form the motion trend trajectory of the preset target.

[0020] S140 combines wind and wave data with motion trend trajectory to correct the simulated position of the preset target and generate the preset trajectory of the preset target.

[0021] S150 converts the pre-trajectory into a pre-pointing quantity corresponding to the acquisition device, and combines it with the ship's attitude data for attitude compensation, forming the turntable control quantity of the acquisition device.

[0022] S160 uses turntable control to control the acquisition device to acquire data from a preset target.

[0023] In practice, marine environmental data is continuously acquired during ship navigation. This data includes at least wind speed, wind direction, wave height, wave direction, and wave period, and may also be recorded in conjunction with visibility, rainfall intensity, sea fog conditions, and ambient light. Wind speed indicates the speed of air movement around the ship relative to the hull or geographic coordinates; wind direction indicates the direction from which the wind blows. Wave height indicates the degree of undulation between wave crests and troughs; wave direction indicates the direction of wave propagation; and wave period indicates the time interval between adjacent wave crests passing the same reference position. This data can be obtained from shipborne anemometers, wave-measuring radar, sea state estimation data from inertial navigation equipment, meteorological interface data, or integrated ship navigation data. To accommodate subsequent wind and wave data generation, an environmental timestamp corresponding to each piece of marine environmental data needs to be recorded during data collection. The environmental timestamp indicates the specific time when the data was collected or updated.

[0024] In actual deployment, anemometers should be installed in an open location above the deck, avoiding mast obstruction, exhaust vent disturbance, and superstructure backflow areas, to ensure that wind speed and direction more closely approximate the actual wind conditions of the sea area where the ship is located. Wave-measuring radar or sea state sensor units should be installed in areas covering the sea surface ahead of and beside the ship to obtain wave height, wave direction, and wave period. When the ship has an integrated navigation system, wave height, wave direction, and wave period estimates can also be read from the integrated navigation system, but the original source identifier must be retained. The original source identifier indicates whether the marine environmental data comes from an independent sensor, the ship's navigation interface, or an external meteorological interface. Before the marine environmental data enters subsequent processing, isolated points that clearly exceed the reasonable range of sea state conditions are removed. For example, if the wind direction suddenly jumps within a short period of time, inconsistent with actual meteorological changes, or if the instantaneous wave height value deviates significantly from multiple preceding and following sampling values, the sampling point is marked as an abnormal environmental point.

[0025] Ship attitude data is acquired synchronously during the operation of the acquisition equipment. This data includes at least roll, pitch, and bow angles, and may also include roll rate, pitch rate, and bow rate. The roll angle represents the angle of heeling or roll about the ship's longitudinal axis; the pitch angle represents the angle of bow-stern heeling about the ship's transverse axis; and the bow angle represents the heading angle of the ship's bow relative to a geographic reference direction. The roll rate, pitch rate, and bow rate indicate how quickly the corresponding attitude angles change over time, and are used to determine whether the ship's rolling motion is intensifying, decreasing, or reversing. Ship attitude data can be obtained from onboard inertial navigation units, attitude sensors, gyrocompasses, or attitude reference units.

[0026] To ensure accurate use of ship attitude data in the turntable control of the data acquisition equipment, a correspondence must be established between the acquired data and the equipment's installation reference after reading. The installation reference is determined by the fixed attitude of the acquisition equipment at its installation location on the ship, including the equipment's installation deflection angles relative to the ship's bow, lateral, and vertical directions. If the acquisition equipment is installed on a deck platform or mast platform, the installation deflection angle is recorded through static calibration after installation and written into the acquisition parameter table. The acquisition parameter table represents the fixed installation relationship of the acquisition equipment on the ship. Subsequent readings of roll, pitch, and bow angles will deduct the installation deflection angle, ensuring that the ship attitude data reflects the actual line-of-sight direction of the acquisition equipment, rather than merely remaining at the overall ship attitude level.

[0027] A preset target refers to a small, high-speed target that needs to be continuously acquired by the acquisition equipment. Preset targets can be unmanned surface vessels, low-altitude small aircraft, or fast-drifting targets on the sea surface. Preset targets can be selected manually on the display screen, or determined by image recognition results, radar target information, or external target indication information. When selected manually, the operator selects a target area in the visible light, low-light, or infrared camera image; the target area is then processed by image tracking to form the preset target. When determined by external target indication information, the azimuth, distance, or latitude and longitude of the external target is first converted into the observation direction of the acquisition equipment, and then the acquisition equipment is driven to search the corresponding field of view. Once the target appears in the image, the preset target is confirmed. After confirmation, a target number is assigned to the preset target, and this target number remains consistent across visible light, low-light, and infrared camera images.

[0028] Historical trajectory data is continuously generated around the same preset target. This data includes at least the target's azimuth, elevation, distance, acquisition channel identifier, target confidence level, and trajectory timestamp at multiple acquisition times. Azimuth represents the angle of the preset target's horizontal rotation relative to the acquisition device; elevation represents the angle of the preset target's vertical pointing direction relative to the acquisition device; and distance represents the spatial interval between the preset target and the acquisition device. The acquisition channel identifier indicates whether the target observation result comes from visible light imaging, low-light imaging, or infrared imaging. Target confidence level indicates the reliability of the target identification result at that acquisition time and can be determined based on the target outline clarity, target area stability, continuity of infrared thermal features, and consistency of positional changes in adjacent frames. The trajectory timestamp indicates the acquisition time corresponding to the target observation result.

[0029] Historical trajectory data can be generated from the image tracking results of the acquisition device. Visible light cameras output the target bounding box, target center, and target sharpness during the day or in good visibility conditions. Low-light cameras output the target bounding box, target center, and the target area with enhanced brightness under low-light conditions. Infrared cameras output the hot target area and hot target center at night or when the target's thermal features are obvious. After obtaining the target center for each acquisition channel, the image coordinates are converted into the preset target's azimuth and elevation by combining the azimuth, elevation, and lens field of view of that acquisition channel. If the acquisition device can receive radar ranging, laser ranging, or external target distance information, the corresponding distance is written into the historical trajectory data at the same acquisition time. For targets whose distance cannot be directly measured, the distance can be estimated based on the target pixel size, known target type size, and lens focal length, and this estimated distance is marked as the estimated distance.

[0030] When switching between multiple acquisition channels, historical trajectory data is still recorded continuously according to the same target number. For example, if a preset target is acquired by a visible light camera during the day, then by a low-light camera at dusk, and finally by an infrared camera at night, each frame of target observation results is written into the historical trajectory data of the same preset target, with the source channel only indicated in the acquisition channel identifier. If both the visible light and infrared cameras acquire target observation results during the switching process, the observation result with higher target confidence is prioritized as the main trajectory point at the current acquisition time, while the other observation result is retained as an auxiliary trajectory point. The main trajectory point represents the target point directly used in subsequent trajectory processing, while the auxiliary trajectory point represents the target point used to verify the continuity of the target and the consistency of the channels.

[0031] Marine environmental data, ship attitude data, and historical trajectory data need to be synchronized in time. Time synchronization can use the local clock of the acquisition equipment, the ship's navigation clock, or a unified time synchronization signal as the time reference. The environmental timestamps of the marine environmental data, the attitude timestamps of the ship attitude data, and the trajectory timestamps of the historical trajectory data are all converted to the same time reference. If the sampling frequencies of the three types of data are different, the trajectory timestamp of the historical trajectory data is used as the alignment reference, and the marine environmental data and ship attitude data with the closest time to each trajectory timestamp are selected. If the time difference exceeds a preset allowable range, interpolation is performed between two adjacent sampling points to form environmental synchronization data and attitude synchronization data corresponding to that trajectory timestamp. Environmental synchronization data represents marine environmental data that has been aligned with historical trajectory data, and attitude synchronization data represents ship attitude data that has been aligned with historical trajectory data.

[0032] After synchronization, a completeness check is performed on the marine environmental data, ship attitude data, and historical trajectory data. This check includes determining if wind speed, wind direction, wave height, wave direction, and wave period are missing; whether roll, pitch, and heading angles are continuously updated; and whether bearing, pitch, distance, and acquisition channel identifiers correspond to the same preset target. If marine environmental data is missing for a short period, data from adjacent short periods with relatively stable changes can be used, but this short period needs to be marked as an environmental compensation segment. If ship attitude data is missing, subsequent attitude compensation data is not generated directly; instead, the system waits for the next valid attitude sampling point or reduces the tracking speed of the acquisition equipment. If historical trajectory data is missing, the missing marker is retained, and transitional target points are subsequently added using motion trend trajectories.

[0033] In one possible implementation, after the marine environmental data enters the processing stage, it is first organized according to the acquisition time, grouping wind speed, wind direction, wave height, wave direction, wave period, visibility, and ambient light into the same environmental record. Wind speed indicates the strength of the sea surface wind field, and wind direction indicates the direction of the wind field relative to the geographic reference direction; wave height indicates the amplitude of sea surface undulation, wave direction indicates the direction of wave propagation, and wave period indicates the time characteristics of wave repetition; visibility and ambient light are used to determine whether environmental conditions such as sea fog, rain, and low nighttime illumination will affect the image observation of preset targets. Each environmental record retains an environmental timestamp, which is used for alignment with ship attitude data and historical trajectory data.

[0034] Since wind speed, wind direction, wave height, and wave direction typically originate from different sensor sources, and the update frequencies of these sources are not entirely consistent, it is necessary to use the control cycle of the data acquisition equipment as a unified time reference to align the marine environmental data. If multiple wind speed or wave height records exist within a given control cycle, the record closest to that control cycle is used. If two consecutive environmental records span the control cycle, linear interpolation is performed according to the time interval to obtain the corresponding wind speed, wind direction, wave height, wave direction, and wave period for that control cycle. Environmental records generated in this way can be directly used in subsequent wind and wave data calculations without causing jumps in wind and wave data due to different sampling intervals from different sensor sources.

[0035] After time alignment, outlier processing is performed on the marine environmental data. If the wind speed changes abruptly within a short period of time, significantly exceeding the actual wind conditions in the area where the ship is located, or if the wind direction experiences a sudden, reverse jump that does not conform to continuous meteorological changes, the corresponding wind speed or direction is marked as a wind field anomaly. If the wave height suddenly becomes much larger than multiple preceding and following sample values, or if the wave period becomes unreasonably small for a short period, the corresponding wave height or wave period is marked as a wave field anomaly. Wind field and wave field anomalies do not directly participate in the generation of wind and wave data; instead, they are corrected by adjacent valid environmental records to prevent single-sensor errors from being amplified into turntable control deviations.

[0036] After correcting the marine environmental data, the wind and wave directions are transformed to the ship's coordinate system. The ship's coordinate system is a coordinate relationship established based on the ship itself, typically with the bow direction as longitudinal, the starboard direction as transverse, and the vertical direction as altitude. Wind and wave directions are often based on geographic north or a heading system, which cannot be directly used to determine the impact of wind and waves on the ship's roll and the motion of the target. Therefore, it is necessary to calculate the relative wind and wave directions in conjunction with the ship's bow angle. Relative wind direction represents the directional relationship of the wind field relative to the ship's bow, and relative wave direction represents the directional relationship of the waves relative to the ship's bow. Downwind, crosswind, and headwind have different effects on the ship's roll and the target's offset; similarly, downwind, crosswind, and headwind have different effects on the stability of the turntable's line of sight. The above transformation preserves these directional differences. The relative wind and wave directions can be determined using the following expressions:

[0037] in, Indicates relative wind direction; Indicates wind direction; Indicates the bow angle of the hull; Indicates relative wave direction; The relative wind direction and relative wave direction can be normalized to a preset angle range after calculation, enabling them to stably represent the directional relationship between the wind field and the wave field relative to the ship's bow. The above expression converts the wind direction and wave direction under the geographical reference to directional quantities under the ship's reference by subtracting the ship's bow angle.

[0038] After obtaining the relative wind direction, the wind speed is decomposed into longitudinal and lateral wind volumes. The longitudinal wind volume represents the wind field component acting along the bow or stern direction, primarily affecting the relative approach or departure of the target along the line of sight. The lateral wind volume represents the wind field component acting along the port and starboard sides of the ship, primarily affecting the drift of the target in the lateral direction of the image, and also exacerbating the line-of-sight shift caused by the ship's roll. The wind volume decomposition expression is:

[0039] in, Indicates longitudinal air volume; Indicates lateral airflow; Indicates wind speed; This indicates relative wind direction. Both longitudinal and lateral wind volumes originate from the same wind speed, but are decomposed into different directions within the ship's coordinate system according to relative wind direction. This decomposition transforms the wind field from a simple strength value into directional data that can be used in ship roll analysis and preset target offset analysis.

[0040] After obtaining the relative wave direction, the wave height is decomposed into longitudinal wave volume and transverse wave volume. Longitudinal wave volume represents the wave undulation effect propagating along the bow or stern direction, primarily corresponding to pitch disturbance; transverse wave volume represents the wave undulation effect propagating along the port and starboard sides of the ship, primarily corresponding to roll disturbance. The wave volume decomposition expression is:

[0041] in, Indicates vertical swell; H represents the horizontal wave volume; H represents the wave height. The relative wave direction is indicated. Longitudinal wave volume reflects the impact of waves on the ship's pitch and roll and the observation of the preset target height, while lateral wave volume reflects the impact of waves on the ship's roll and the lateral shift of the preset target observation. Through this processing, wave height and wave direction are synthesized into directional wave field data that can be used for subsequent attitude compensation and pre-trajectory correction.

[0042] Wave period is used to describe the rate of change of wave undulation. When the wave period is short, the hull experiences faster changes in undulation disturbance, and the preset target switches more frequently between wave crests, troughs, and spray obstruction. When the wave period is long, wind and wave disturbances change relatively slowly, and the turntable control can use a relatively gentle compensation method. To incorporate the wave period into the wind and wave data, wave variation intensity can be generated based on wave height and wave period. Wave variation intensity represents the trend of the impact of wave undulation on the acquisition line of sight per unit time, and can be determined by the following formula:

[0043] in, H represents wave intensity; H represents wave height. The wave period is represented by ε; ε represents the stability term, used to prevent calculation anomalies caused by missing or near-zero wave periods. This expression describes the intensity of wave fluctuations through the relationship between wave height and wave period. The greater the wave height and the shorter the wave period, the higher the wave intensity, and the more significant the subsequent corrections for ship roll and preset target motion need to be.

[0044] After generating longitudinal wind volume, transverse wind volume, longitudinal wave volume, transverse wave volume, and wave variation intensity, the hull roll disturbance is further generated. The hull roll disturbance is used to characterize the impact of wind and waves on the ship's roll, pitch, and heading stability. Transverse wind volume and transverse wave volume primarily participate in roll disturbance calculations, while longitudinal wave volume primarily participates in pitch disturbance calculations. Longitudinal and transverse wind volumes also affect heading disturbance assessment. The hull roll disturbance does not directly replace hull attitude data but is used to explain the external causes of hull attitude changes under current wind and wave conditions and to provide lead time for subsequent attitude compensation. The hull roll disturbance can be determined using the following expression:

[0045] in, Indicates the amount of roll disturbance; Indicates the amount of pitch disturbance; Indicates lateral airflow; Indicates horizontal swell; Indicates vertical swell; Indicates wave intensity; This represents the transverse air volume coefficient, with a value ranging from 0 to 1; This represents the horizontal sway coefficient, with a value ranging from 0 to 1; This represents the longitudinal wave volume coefficient, with a value ranging from 0 to 1; This represents the wave variation coefficient, with a value ranging from 0 to 1. These coefficients are used to limit the proportion of the corresponding wind and wave components involved in the ship's roll disturbance. The values ​​are set to 0 to 1 because different ship types, loading conditions, and the installation location of the data acquisition equipment respond differently to wind and waves, and the coefficients should not exceed the actual principal disturbance quantities. This expression synthesizes the wind and wave components into roll and pitch disturbance quantities, enabling the wind and wave data to express the impact on the ship's roll.

[0046] To assess the impact of a pre-defined target's motion, target offset perturbation needs to be derived from wind and wave data. When the pre-defined target is a rapidly drifting target on the sea surface, its motion is easily offset by wind and wave fields. When the pre-defined target is an unmanned surface vessel (USV), although it has active propulsion, it may still experience short-term offsets during high-speed turns, wave impacts, or in transverse wave areas. When the pre-defined target is a low-altitude small aircraft, the wind field has a more significant impact on its lateral drift, while the wave field mainly affects the image background and the target's occlusion status. The target offset perturbation is used to represent the environmental offset trend of the pre-defined target's future position caused by wind and waves.

[0047] The target offset disturbance can be adjusted according to the preset target type. For rapidly drifting targets on the sea surface, lateral wind volume, longitudinal wind volume, lateral wave volume, and longitudinal wave volume are all included in the calculation; for unmanned surface vessels, the participation ratio of wind volume and wave volume is mainly determined based on the motion trend trajectory to determine whether wind and waves cause continuous offset; for low-altitude small aircraft, longitudinal and lateral wind volume are mainly used to generate the target offset disturbance. The target offset disturbance can be expressed as:

[0048] in, This represents the target offset disturbance; this expression converts the wind field and wave field into a three-dimensional offset relationship, which is used to subsequently correct the simulated position of the preset target.

[0049] When visibility, rainfall intensity, sea fog conditions, and ambient light are involved in the processing, scene impact markers can be generated in the wind and wave data. Scene impact markers indicate the influence of current wind, wave, and weather conditions on the stability of the acquisition channel. For example, when visibility decreases and sea fog intensifies, scene impact markers indicate that visible light cameras are significantly affected by fog; when ambient light decreases, scene impact markers indicate that low-light or infrared cameras are more suitable for subsequent acquisition; when rainfall intensity increases and lateral wave volume is high, scene impact markers indicate that rain curtains, waves, and water films may appear in the image. Scene impact markers are not directly used as the acquired image results, but are provided along with the wind and wave data for subsequent acquisition channel selection and pre-trajectory reliability verification.

[0050] The final generated wind and wave data can include wind field components, wave field components, wave intensity, ship roll disturbance, target offset disturbance, and scene influence markers. The wind field components include longitudinal and lateral wind volumes; the wave field components include longitudinal and lateral wave volumes; the ship roll disturbance includes roll and pitch disturbances; the target offset disturbance is used to characterize the environmental offset trend of the preset target under the influence of wind and waves; and the scene influence markers are used to characterize the impact of current wind, wave, and weather conditions on the acquisition channel.

[0051] In one possible implementation, when processing historical trajectory data, the target observation records formed by the acquisition device during continuous acquisition are first organized. Each target observation record includes at least the acquisition time, target number, azimuth, elevation, distance, acquisition channel identifier, and target confidence level. The acquisition time indicates the time when the target observation record was formed; the target number indicates the unified identity of the same preset target under different acquisition channels and at different acquisition times; the azimuth indicates the angle of the preset target relative to the horizontal rotation direction of the acquisition device; the elevation indicates the angle of the preset target relative to the vertical pointing direction of the acquisition device; the distance indicates the spatial interval between the preset target and the acquisition device; the acquisition channel identifier indicates that the target observation record originates from visible light imaging, low-light imaging, or infrared imaging; and the target confidence level indicates the degree of confidence that the current target observation record belongs to the preset target, which is usually determined by the completeness of the target outline, the stability of the target center, the continuity of thermal features, and the consistency of positional changes in adjacent acquisition frames.

[0052] When organizing historical trajectory data, target observation records are rearranged according to the acquisition time from earliest to latest. If multiple acquisition channels simultaneously acquire target observation records near the same acquisition time, they are merged based on target number, azimuth proximity, elevation proximity, distance proximity, and target confidence level. The merged data forms a target sequence, where each target point corresponds to the same preset target, and the acquisition channel identifier corresponding to that target point is retained. In this way, when data is acquired by visible light cameras during the day, followed by low-light cameras at dusk, and then by infrared cameras at night, the target sequence maintains a continuous relationship of change with the same preset target.

[0053] When multiple target observation records exist near the same acquisition time, the angle difference and distance difference between the target observation records can be calculated first. The angle difference is used to determine whether the target centers obtained from different acquisition channels fall in similar pointing directions, and the distance difference is used to determine whether the target distances corresponding to different acquisition channels are consistent. The angle difference can be determined according to the following formula:

[0054] in, This represents the angular difference between two target observation records; Indicates the location of a target in an observation record; Indicates the location of another target in the observation record; This represents the elevation in a target observation record; This represents the elevation in another target observation record. This formula combines the azimuth difference and elevation difference into the same angular deviation, used to determine whether two target observation records point to the same preset target.

[0055] The angle merging threshold can be set from 0.05° to 0.3°. This range is related to the field of view during shipborne telephoto acquisition, the accuracy of the turntable encoder, and the short-term displacement of high-speed small targets. If the threshold is too small, slight line-of-sight deviations in the acquisition channel may prevent the same preset target from being merged; if the threshold is too large, sea surface spray, hot spots, or other small targets may be incorrectly classified into the same preset target. The distance merging threshold can be set from 2% to 8% of the target distance. This range is used to account for the error differences between radar ranging, image ranging, and laser ranging. Target observation records with a target confidence score below 0.35 are not directly used as primary target points; target observation records with a target confidence score between 0.35 and 0.6 are used as auxiliary target points; and target observation records with a target confidence score above 0.6 can be used as primary target points. The rationale behind setting these ranges is that, under conditions of rain and fog at sea, low light at night, and infrared hot spot interference, single-frame recognition results should not be discarded too early, but low-confidence records should not be directly used in motion trend calculations.

[0056] After the target sequence is formed, the target points in the sequence are preprocessed. Preprocessing includes outlier removal, missing point insertion, channel bias correction, and confidence level update. Outliers are target points that change drastically relative to preceding and following target points, such as sudden reversals in azimuth, sudden elevation changes, or sudden large changes in distance, even though the preceding and following target points maintain a continuous movement trend. Missing points are locations where there is a short period of no target recording due to obstruction by waves, rain, sea fog, or switching of acquisition channels. Channel bias refers to the fixed bias introduced by visible light imaging, low-light imaging, and infrared imaging due to differences in installation location, line-of-sight calibration, and imaging field of view. Confidence level update is used to feed back the continuity of target points into the target confidence level, giving higher confidence to continuous and stable target points, and maintaining lower confidence for target points with abrupt changes or obstruction.

[0057] When removing outliers, velocity changes between adjacent target points can be used for judgment. First, estimate short-term motion changes based on the changes in azimuth, pitch, and distance of adjacent target points in the target sequence. If the change between a target point and the preceding target point is significantly greater than the average change of consecutive segments, and there is no continuous motion relationship between this target point and the following target point, then this target point is marked as an outlier. The outlier determination can be expressed using the following formula:

[0058] in, This represents the jump value of the target point at time t; This represents the initial spatial position of the target point at time t; This indicates the preliminary spatial position of the target point at the previous moment; This indicates the preliminary spatial position of the target point at the next moment; This represents the time interval from the previous moment to moment t. This represents the time interval from time t to the next time. This represents a stability term, used to avoid calculation anomalies caused by excessively small time intervals. This formula uses the rate of spatial change between the target point and its preceding and following points to determine whether a sudden jump exists at the target point.

[0059] The anomaly threshold can be set to 2 to 4 times the median velocity of the target sequence's nearest neighbors. This is because high-speed, small targets are inherently allowed to accelerate and turn; a threshold that is too low will mistakenly delete genuine maneuvering points, while a threshold that is too high will retain obvious jumps caused by splashes, heat spots, or misidentification. When the preset target is a low-altitude small aircraft, a higher anomaly threshold can be used; when the preset target is a rapidly drifting target on the sea surface, a lower anomaly threshold can be used; and when the preset target is an unmanned surface vessel, an intermediate value should be selected based on its historical velocity variation range.

[0060] When imputing missing points, if a target sequence is missing within a short period, and the target points before and after the missing point belong to the same preset target, interpolation is performed according to the motion trend before and after the missing point. Short-term missing points are typically set to 1 to 5 frames, or 0.1 to 1.0 seconds. This range is chosen because ocean wave obstruction and channel switching usually last for short periods; continuing interpolation beyond this range might mistakenly identify lost targets as still existing. Missing points can be imputed using the following formula:

[0061] in, Indicates the location of the missing point to be filled in; Indicates the location of the valid target point before it was missing; This indicates the location of the valid target point after the missing data; t represents the acquisition time corresponding to the missing point. Indicates the time of collection of valid target points before they became missing; This indicates the acquisition time of the valid target points after the missing data. This formula fills in the target positions according to the time ratio between the valid target points before and after the missing data, so that the target sequence remains continuous in short-term occlusion sections.

[0062] During channel deviation correction, the fixed line-of-sight deviation between visible light imaging, low-light imaging, and infrared imaging is recorded in the line-of-sight calibration table. The line-of-sight calibration table records the azimuth and pitch correction values ​​of different acquisition channels relative to a unified line of sight. If a target point originates from low-light or infrared imaging, the corresponding correction value is read according to its acquisition channel identifier, and the azimuth and pitch of that target point are corrected. Channel correction can be determined using the following formula:

[0063] in, Indicates the corrected orientation; Indicates the orientation before correction; This indicates the azimuth correction value corresponding to the acquisition channel identifier; Indicates the corrected pitch; Indicates the pitch before correction; This represents the pitch correction value corresponding to the acquisition channel identifier. This formula uses a line-of-sight calibration table to unify target points under different acquisition channels to the same line-of-sight reference, reducing trajectory breaks caused by channel switching.

[0064] After outlier removal, missing point insertion, and channel deviation correction, a corrected target sequence is formed. Each target point in the corrected target sequence includes the corrected azimuth, corrected elevation, distance, acquisition channel identifier, target confidence level, and acquisition time. If a target point is an insertion point, its insertion identifier is retained in the corrected target sequence; if a target point has undergone channel deviation correction, its acquisition channel identifier and correction source are retained to facilitate subsequent determination of whether the target point was affected by acquisition channel switching.

[0065] Next, the bearing, pitch, and range in the corrected target sequence are converted into the target position in the ship's coordinate system. The ship's coordinate system uses the ship as a spatial reference, facilitating the correlation between the target position and ship attitude data, wind and wave data, and turntable control variables. During the conversion, bearing represents the horizontal deflection direction, pitch represents the vertical deflection direction, and range represents the spatial length from the acquisition device to the preset target. The target position can be determined using the following formula:

[0066] in, Indicates the target position at time t; This represents the distance between the preset target and the acquisition device at time t; Indicates the corrected orientation at time t; This represents the corrected elevation at time t. This formula converts the spherical observation relationship into a spatial coordinate relationship. The lateral and longitudinal positions are determined by distance, elevation, and azimuth, while the altitude position is determined by distance and elevation.

[0067] If the acquisition device has an installation offset relative to the origin of the ship's coordinate system, its installation coordinates need to be superimposed onto the target position. The installation coordinates represent the position of the optical center of the acquisition device relative to the origin of the ship's coordinate system. The superimposed target position can be determined using the following formula:

[0068] in, C represents the target position in the ship's coordinate system at time t; C represents the installation coordinates of the data acquisition equipment. This represents the target position relative to the acquisition device at time t. This formula transforms the target position from the local reference of the acquisition device to the reference of the ship's coordinate system, which facilitates subsequent turntable control by combining the ship's attitude data.

[0069] After obtaining the continuous target positions, the target positions are smoothed to form a smooth position sequence. Smoothing is used to suppress short-term jumps caused by ship swaying, camera shake, long-distance ranging errors, and single-frame image recognition errors, while preserving the true turning and acceleration trends of the preset targets. Smoothing can employ a recursive smoothing method:

[0070] in, Indicates the smoothed position at time t; This represents the target position in the ship's coordinate system at time t; Indicates the smoothed position at the previous moment; This represents the smoothing coefficient, with a value ranging from 0.4 to 0.8. The smoothing coefficient determines the proportion of the current target position in the smoothing result. A value of 0.4 to 0.8 is chosen because high-speed, small targets need to retain their current motion changes; a smoothing coefficient that is too low will cause the target trajectory to lag, while a smoothing coefficient that is too high will cause more noise from a single frame to enter the smoothed area. When the preset target speed is relatively high, the smoothing coefficient can be set to 0.65 to 0.8; when the target is significantly obscured by waves, the smoothing coefficient can be set to 0.4 to 0.6.

[0071] After generating a smooth position sequence, the target velocity is determined based on this sequence. Target velocity represents the rate of change of the target's position and direction of motion between adjacent acquisition times. The target velocity can be determined using the following formula:

[0072] in, Represents the target velocity at time t; Indicates the smoothed position at time t; Indicates the smoothed position at the previous moment; This represents the time interval between adjacent acquisition moments. This formula uses bit removal between adjacent smooth positions to obtain the target velocity over time intervals, reflecting the current main direction of movement and speed of the preset target.

[0073] Target acceleration represents the change of target velocity over time and is used to determine whether a preset target is accelerating, decelerating, or maneuvering. Target acceleration can be determined by the following formula:

[0074] in, Represents the target acceleration at time t; Represents the target velocity at time t; Indicates the target velocity at the previous moment; This represents the time interval between adjacent acquisition moments. This formula determines the target acceleration by considering the changes in the target's velocity between consecutive targets, enabling subsequent pre-trajectory generation to take into account the velocity changes of the preset target, rather than simply extrapolating based on a uniform velocity.

[0075] Steering trend is used to determine whether the direction of motion of a preset target continues to change. The steering angle can be determined by the angle between the velocity directions of adjacent targets, and then multiple consecutive steering angles can be used to determine whether a stable steering has been formed. The steering angle can be determined according to the following formula:

[0076] in, This represents the turning angle at time t; Represents the target velocity at time t; Indicates the target velocity at the previous moment; This represents the stabilizing term, used to avoid an abnormal denominator when the target speed is too low. The formula reflects the degree of change in the preset target's motion direction through the angle between consecutive target speeds; a larger turning angle indicates a more significant change in the preset target's motion direction.

[0077] The turning trend can be set to three states: stable straight-ahead, slight turning, and continuous turning. If the turning angle is less than 3° for 3 to 5 consecutive acquisition time periods, the turning trend is marked as stable straight-ahead; if the turning angle is between 3° and 10° for 3 to 5 consecutive acquisition time periods, the turning trend is marked as slight turning; if the turning angle is greater than 10° for 3 to 5 consecutive acquisition time periods, and the turning direction is consistent, the turning trend is marked as continuous turning. The reason for setting this range is that long-distance shipborne acquisition may have small measurement jitters. Within 3°, it can usually be classified as observation fluctuation; 3° to 10° can reflect slow changes in the target's movement direction; when it is above 10° and occurs continuously, it usually indicates that the high-speed small target has undergone significant maneuvering. The number of consecutive acquisition time periods is set to 3 to 5 to avoid misjudgment of the turning trend due to single-frame jumps, while maintaining the response speed to changes in the maneuvering of high-speed small targets.

[0078] To form a complete motion trend trajectory from the target velocity, acceleration, and turning trend, the smoothed position, target velocity, acceleration, turning trend, acquisition channel identifier, and target confidence level need to be correlated according to the acquisition time. Each acquisition time corresponds to a motion trend point, representing the motion state of the preset target at that acquisition time. Multiple motion trend points are connected in chronological order to form the motion trend trajectory. The motion trend trajectory records not only the position traversed by the preset target but also its direction of movement, speed, velocity changes, turning state, and acquisition source.

[0079] In one specific embodiment, the preset target is a high-speed unmanned surface vessel (USV) on the sea surface. During the day, visible light cameras continuously acquire the USV's azimuth, pitch, and distance. Upon entering the reflective area of ​​waves, the target confidence of some target points decreases, and a single frame of azimuth jump occurs. In processing, a target sequence is first formed according to the acquisition time. Then, the azimuth jump point is identified as an anomaly and removed based on the jump value. Subsequently, valid target points before and after the anomaly point are used to fill in the missing positions, forming a corrected target sequence. Next, the azimuth, pitch, and distance in the corrected target sequence are converted into target positions in the ship's coordinate system, and a smoothing coefficient of 0.65 is used for recursive smoothing to form a smoothed position sequence. The target velocity is calculated based on adjacent smoothed positions, and the target acceleration is calculated based on the continuous target velocities. Finally, the angle between the directions of the continuous target velocities is used to determine whether the USV is continuously turning. If the turning angle at four consecutive acquisition times is greater than 10° and the direction is consistent, the segment is marked as a continuously turning segment.

[0080] In one possible implementation, when generating the pre-trajectory, a segment of motion trend points closest to the current acquisition time is first extracted from the motion trend trajectory. A motion trend point refers to a preset target motion state recorded at the same acquisition time, typically including smooth position, target velocity, target acceleration, turning trend, acquisition channel identifier, and target confidence level. It is not advisable to use only the most recent motion trend point, nor is it advisable to use motion trend points from an excessively long time period; the former is easily affected by single-frame disturbances, and the latter will carry over earlier motion states into the current extrapolation. The extraction window can be set to the most recent 3 to 8 motion trend points, or to motion trend points from the most recent 0.3 to 2 seconds. The reason for this range is that the maneuvering changes of unmanned surface vessels, low-altitude small aircraft, and rapidly drifting targets on the sea surface are rapid; fewer than 3 motion trend points make it difficult to determine the turning trend, and historical states exceeding 2 seconds are prone to inconsistencies with the current motion.

[0081] The current smoothed position, target velocity, target acceleration, and turning trend are determined from the captured motion trend points. The current smoothed position is generally taken from the smoothed position among the most recent valid motion trend points; the target velocity can be the average or weighted velocity of multiple recent motion trend points; the target acceleration can be the average or weighted acceleration of multiple recent motion trend points; the turning trend is determined by the velocity direction changes of multiple recent motion trend points. If a target point is added due to short-term occlusion among the recent motion trend points, this target point does not independently determine the target velocity and target acceleration, but only participates in maintaining continuity. The acquisition channel identifier is also retained for subsequent judgment on whether the corrected position is affected by acquisition channel switching, low-light noise, rain / fog obstruction, or infrared hotspots. The weighted determination of the target velocity can be performed according to the following formula:

[0082] in, This indicates the target velocity at the current acquisition moment; This represents the velocity corresponding to the q-th trend point forward from the current acquisition time; This represents the confidence factor corresponding to the q-th motion trend point forward from the current acquisition time; n represents the number of motion trend points involved in the calculation. The confidence factor can be jointly determined by the target confidence level, whether it is a supplementary entry point, and the stability of the acquisition channel. Its value range is determined by the measured image quality, target continuity, and channel calibration error. This formula improves the contribution of stable motion trend points to the target velocity through the confidence factor, and reduces the influence of low-confidence target points and supplementary entry points on the target velocity. The target acceleration can be determined in the same way:

[0083] in, This represents the target acceleration at the current acquisition moment; This represents the acceleration corresponding to the q-th motion trend point forward from the current acquisition time; This represents the confidence factor corresponding to the q-th motion trend point forward from the current acquisition time; n represents the number of motion trend points involved in the calculation. This formula uses the confidence factor of the target velocity to ensure that the target acceleration does not fluctuate drastically due to a single abnormal motion trend point.

[0084] After determining the target velocity and target acceleration, a corrected velocity is generated based on the turning trend. When the turning trend is stable straight-line movement, the corrected velocity remains along the target velocity direction; when the turning trend is a slight turn, a small angle correction is made to the target velocity direction; when the turning trend is a continuous turn, the target velocity direction is corrected based on the cumulative trend of the turning directions and turning angles of the most recent multiple motion trend points. The corrected velocity represents the amount of velocity involved in the extrapolation of the simulated position, retaining both the current speed of the preset target and reflecting its turning state. If the turning direction is consistent across multiple consecutive motion trend points, it indicates that the preset target is forming an effective maneuver, and the corrected velocity needs to rotate towards that turning direction; if the turning direction changes back and forth, it is treated as stable straight-line movement or a slight turn. The corrected velocity can be determined using the following formula:

[0085] in, This indicates the correction speed corresponding to the current acquisition time; This indicates the target velocity at the current acquisition moment; This indicates the rotational relationship formed by the modified steering angle; This represents the corrected steering angle corresponding to the current acquisition moment. The corrected steering angle is determined based on the steering angles and directions of the most recent multiple motion trend points, and its limits can be determined through sea trial records, target type, and acquisition frame rate. The function of this formula is to convert the steering trend into a velocity direction correction, so that subsequent simulated positions are no longer simply extrapolated along a straight line.

[0086] Based on the smoothed position, corrected velocity, and target acceleration, the simulated position of the preset target after the prediction time interval is extrapolated. The prediction time interval represents the time difference from the current acquisition moment to the future acquisition moment, and can be determined according to the control cycle of the acquisition equipment, the response time of the turntable mechanism, and the speed of the preset target. For high-speed, small targets, the prediction time interval is usually set to cover the period before the next turntable action is completed; if the prediction time interval is too short, the turntable mechanism cannot point to the future position in advance; if the prediction time interval is too long, the motion extrapolation error will increase significantly. The prediction time interval can be set to one or more control cycles, and a single prediction time interval is usually matched with the image frame cycle, azimuth motor response time, pitch motor response time, and fine-tuning frame response time. The extrapolation expression for the simulated position is:

[0087] in, Indicates the simulated position after the prediction time interval; Indicates the smoothing position corresponding to the current acquisition time; This indicates the correction speed corresponding to the current acquisition time; This represents the target acceleration at the current acquisition moment; Indicates the prediction time interval; This represents the acceleration coefficient. The range of the acceleration coefficient is determined based on the measured fluctuations in target acceleration, target type, acquisition frame rate, and turntable response hysteresis. When the target acceleration is continuously stable, the participation of the acceleration coefficient is increased; when the target acceleration is caused by a single-frame jump, the participation of the acceleration coefficient is decreased. This formula is expanded from the conventional displacement extrapolation relationship, with the current smooth position as the extrapolation starting point. The corrected velocity corresponds to the first-order displacement term, and the target acceleration corresponds to the second-order displacement term. The acceleration coefficient limits the amplification of the simulated position by short-time measurement noise.

[0088] After establishing the simulated position, wind disturbance vector, wave disturbance vector, and wave period are extracted from the wind and wave data. The wind disturbance vector represents the direction and intensity of the wind field's offset from the preset target in the ship's coordinate system, derived from the conversion between wind speed, wind direction, and ship heading. The wave disturbance vector represents the impact of ocean waves on the preset target position, target obstruction, and ship roll in the ship's coordinate system, derived from the conversion between wave height, wave direction, and ship heading. The wave period represents the rate of change of ocean wave fluctuations and is used to determine the intensity of the wave disturbance vector's change within the prediction time interval. The wind and wave disturbance vectors need to be aligned with the acquisition time of the current motion trend point. If the wind and wave data sampling time differs from the motion trend point acquisition time, the most recent time point or interpolation method is used to obtain the wind disturbance vector, wave disturbance vector, and wave period corresponding to the current acquisition time.

[0089] The wind and wave correction is generated based on the wind disturbance vector, wave disturbance vector, and wave period. For unmanned surface vessels (USVs), the wind and wave disturbance vectors primarily correct for lateral offset and wave impact offset during high-speed motion; for rapidly drifting targets on the sea surface, the wind and wave disturbance vectors directly reflect their drift trend; for low-altitude small aircraft, the wind disturbance vector primarily corrects for lateral flight trajectory offset, while the wave disturbance vector mainly participates in the judgment of sea surface background disturbance and short-term occlusion. The expression for the wind and wave correction is:

[0090] in, This indicates the wind and wave correction amount after the prediction time interval; This represents the wind disturbance vector corresponding to the current acquisition moment; This represents the wave disturbance vector corresponding to the current acquisition time; This indicates the wave period corresponding to the current data collection moment; Indicates the prediction time interval; Indicates a stable term; Indicates the wind disturbance coefficient; This represents the wave disturbance coefficient. The range of the wind disturbance coefficient is determined based on the measured wind field offset, the preset target type, and the calibration results of the target's windward area; the range of the wave disturbance coefficient is determined based on the measured wave height response, the preset target's draft or flight altitude, the ship's roll response, and historical acquisition errors. This formula allows the wind disturbance effect to accumulate over the prediction time interval and makes the wave disturbance effect constrained by the wave period. The shorter the wave period, the more obvious the change in wave surface state within the same prediction time interval, and the stronger the contribution of the wave disturbance term to the correction.

[0091] The wind and wave correction is then superimposed on the simulated position to form the corrected position. The corrected position represents the future position after adding wind and wave disturbances to the extrapolated result of the preset target's own motion. The expression for the corrected position is:

[0092] in, Indicates the correction position after the prediction time interval; Indicates the simulated position after the prediction time interval; This represents the wind and wave correction amount after the prediction time interval. This formula combines the active motion extrapolation result generated by the motion trend trajectory with the environmental offset result generated by the wind and wave data, so that the correction position can simultaneously reflect the motion inertia, maneuver changes and marine environmental disturbances of the preset target.

[0093] After obtaining the corrected position, a confidence check is performed on the corrected position in conjunction with the acquisition channel identifier. The confidence check is used to determine whether the corrected position may be affected by acquisition channel switching, rain and fog obstruction, low-light noise, infrared hotspots, or wave reflections. If the acquisition channel identifier shows that the recent motion trend points mainly come from visible light cameras, and wind and wave data or environmental records show decreased visibility, the confidence level of corrected positions with large abrupt changes in the lateral or distance directions is reduced. If the recent motion trend points mainly come from low-light cameras, and the ambient illumination is low, the corrected position offset caused by target contour instability is converged. If the recent motion trend points mainly come from infrared cameras, and there are many hotspots on the sea surface, the confidence level of corrected positions that deviate too much from the existing target velocity direction is reduced.

[0094] Confidence checks can generate location confidence scores. Location confidence scores, used to represent the reliability of a corrected location as a pre-trajectory point, can be determined using the following formula:

[0095] in, This indicates the confidence level of the location after the prediction time interval; This indicates the target confidence level at the current data acquisition moment; This indicates the channel stability at the current acquisition moment; This represents the environmental stability at the current acquisition moment. Target confidence is inherited from the motion trend point, channel stability is determined by the acquisition channel identifier, line-of-sight calibration error, and channel switching status, and environmental stability is determined by visibility, wind and wave disturbance strength, and image occlusion status. This formula uses a product relationship because when any factor—target recognition instability, acquisition channel instability, or environmental instability—decreases, the reliability of the corrected position as a pre-trajectory point will decrease.

[0096] When the confidence level of a location falls below a preset requirement, a convergence process is performed on the corrected location. This convergence process does not discard the corrected location; instead, it moves it closer to the simulated location or the previous reliable corrected location. If the confidence reduction is mainly due to sudden changes in wind and wave data, convergence is directed towards the simulated location; if the confidence reduction is mainly due to a switch in the acquisition channel, convergence is directed towards the previous reliable corrected location; if the confidence reduction is mainly due to a decrease in the target confidence level, the number of subsequent prediction time intervals is reduced, thus decreasing the number of long-term predicted trajectory points. The convergence process can be performed according to the following formula:

[0097] in, Indicates the correction position after convergence; Indicates the correction position before convergence; Indicates a reference position; This represents the convergence coefficient. The reference position can be the simulated position or the previous reliable correction position. The range of the convergence coefficient is determined based on the position confidence, the acquisition channel switching status, and the stability of the wind and wave data; the higher the position confidence, the more the convergence coefficient is biased towards the correction position before convergence, and the lower the position confidence, the more the convergence coefficient is biased towards the reference position. This formula limits the abrupt change amplitude of unreliable correction positions through linear convergence, ensuring that the pre-trajectory remains continuous during occlusion, hot spots, or channel switching.

[0098] When generating multiple correction positions sequentially according to multiple prediction time intervals, near-end prediction points, mid-end prediction points, and far-end prediction points can be set. Near-end prediction points are used for advance pointing of the turntable in the next control cycle, mid-end prediction points are used for continuous smoothing of the turntable control input, and far-end prediction points are used to determine whether the preset target is about to leave the current field of view. Multiple prediction time intervals can be set at equal intervals according to the control cycle, or at non-equal intervals according to the turntable response time and the preset target speed. For faster preset targets, the near-end prediction points are spaced closer together; for preset targets with more stable speeds, the mid-end and far-end prediction points can be spaced further apart. The number of prediction time intervals can be set to 3 to 6. This range can balance the lead time of turntable control and motion prediction error. Too few intervals are not conducive to continuous pre-aiming, while too many intervals can easily lead to long-term prediction deviations. Multiple correction positions are connected in chronological order to form a pre-trajectory. The pre-trajectory can be represented as:

[0099] in, This indicates the pre-trajectory corresponding to the current acquisition time; This indicates the correction position after convergence corresponding to the first prediction time interval; This indicates the corrected position after convergence corresponding to the second prediction time interval; This indicates the corrected position after convergence corresponding to the nth prediction time interval; , to This represents multiple prediction time intervals arranged from nearest to farthest. This formula organizes multiple future locations into a continuous path, allowing the reading of pre-trajectory points corresponding to the current control cycle from the pre-trajectory and converting them into pre-pointing values ​​for the acquisition device.

[0100] In one specific embodiment, the preset target is a high-speed unmanned surface vessel (USV). The motion trend trajectory shows that the USV has been continuously moving towards the right front of the vessel in the last five motion trend points, and shows a tendency to turn left. The data acquisition device reads the last five motion trend points, determines the current smoothed position, target velocity, target acceleration, and turning trend, and then generates a corrected velocity according to the turning trend. Subsequently, the simulated position is extrapolated at three prediction time intervals of 0.2 seconds, 0.4 seconds, and 0.6 seconds. Wind and wave data generated from marine environmental data indicates that the current state is a transverse wave state, with a large transverse wave disturbance vector and a short wave period. Therefore, the wind and wave correction causes the simulated position of the USV to be appropriately shifted in the direction of the transverse wave influence. Since the current acquisition channel identifier shows that the target point mainly comes from visible light cameras, and visibility is beginning to decrease, confidence verification and convergence processing are performed on the corrected positions with large offsets.

[0101] In one possible implementation, refer to Figure 2During the assembly of the turntable mechanism, the hull connector is first fixed at the hull mounting location. The hull mounting location is typically chosen on the upper deck, mast platform, or a side platform with minimal obstruction of the view. The mounting surface needs to be leveled and its rigidity checked to prevent localized hull vibrations from being directly amplified to the turntable mechanism. A reinforcing base plate, vibration damping pads, and a waterproof sealing ring can be installed at the bottom of the hull connector. The reinforcing base plate improves the stress stability of the mounting area, the vibration damping pads reduce high-frequency vibrations from the hull engine, high-speed navigation impacts, and wave impacts, and the waterproof sealing ring prevents rainwater, salt spray, and deck water from entering the bottom of the turntable mechanism. Positioning pin holes and fastening holes are pre-drilled on the hull connector. After installation, using the bow, lateral, and vertical directions of the hull as calibration references, the installation deflection angle of the turntable mechanism relative to the hull coordinate system is recorded. This installation deflection angle is subsequently used to convert the hull attitude data into the actual pointing relationship of the turntable mechanism. The original turntable in the document adopted a combination structure of bearing cabin, pitch support and azimuth seat, and eliminated the disturbance of the ship's sway to the optical load through gyroscope, motor and encoder. The ship connecting seat is to supplement the above ship-borne turntable structure to install the reference and wind and wave conditions bearing relationship.

[0102] The azimuth mount is installed above the hull connecting seat, and the azimuth mount and the hull connecting seat are vertically supported by a slewing bearing or upper and lower bearings. The azimuth shaft system adopts a hollow vertical shaft structure. The azimuth motor drives the azimuth shaft system to rotate through a gear pair, synchronous belt, or direct drive structure. The azimuth encoder is installed coaxially with the azimuth shaft system to provide feedback on the actual azimuth angle of the azimuth shaft system. The azimuth angle represents the rotation angle of the turntable mechanism in the horizontal plane relative to the installation reference. Slip rings are arranged inside the azimuth shaft system or at the lower end of the azimuth shaft system. Power cables, video cables, control cables, and environmental data cables are transmitted through the slip rings, preventing cable entanglement during continuous rotation of the azimuth shaft system. The azimuth mount housing adopts a labyrinth seal combined with a friction seal structure. The bottom of the azimuth mount is equipped with a drainage groove and a salt spray protective coating to ensure that the ship can maintain stable rotation under rain, salt spray, and wave splash conditions. The document describes a directional base that uses components such as a directional motor, encoder, directional shaft system, slip ring, and bearings, and can achieve rotation along the directional axis. This implementation structure further introduces continuous transmission and installation reference adaptation on this basis.

[0103] A roll compensation ring is positioned between the azimuth mount and the pitch support. The rotation axis of the roll compensation ring is aligned with the line-of-sight of the support cabin, and it is used to compensate for the tilt of the view and lateral slippage of the line-of-sight caused by the ship's roll. The roll angle represents the angle of left or right tilt of the ship about its longitudinal axis, and the roll rate represents how quickly the roll angle changes over time. A roll motor is fixed to the upper part of the azimuth mount or the outside of the roll compensation ring, and its output is connected to the roll compensation ring. A roll encoder is positioned on the rotation axis of the roll compensation ring to provide feedback on the roll compensation angle. A roll gyroscope is positioned close to the roll compensation ring to sense the roll rate. The roll compensation angle represents the compensation angle produced by the roll compensation ring relative to the azimuth mount. When the ship rolls to the left or right, the roll motor drives the roll compensation ring to rotate in the opposite direction, maintaining a relatively stable line-of-sight attitude for the pitch support and the support cabin mounted on the roll compensation ring. Unlike ordinary camera turntables that rely solely on electronic image stabilization or two-axis gimbal tracking, this structure first weakens the ship's roll at the mechanical level, and then the pitch support and fine-tuning frame complete the subsequent pointing correction.

[0104] The pitch support is mounted on the roll compensation ring. The pitch support employs a double-sided support structure, with an opening between the two sides for installing the carrier chamber. The pitch axis system runs through both sides of the carrier chamber. The pitch motor drives the pitch axis system to rotate via its motor shaft. The pitch encoder is coaxially arranged with the pitch axis system to provide feedback on the actual pitch angle. The pitch angle represents the vertical rotation angle of the carrier chamber relative to the horizontal direction. Pitch limiters are located inside the pitch support or at the end of the pitch axis system to limit the maximum upward and downward pitch angles of the carrier chamber, preventing it from impacting the support due to control malfunctions under high wind and wave conditions. The pitch support can adopt a U-shaped support design, with the motor, encoder, and transmission components arranged within the two side arms. Sealed cavities and drainage cavities are provided inside the support arms. The pitch support in this invention mainly consists of a support, motor shaft, resolver shaft, motor, azimuth gyroscope, encoder, and servo control board. The pitch axis system is supported by bearings. Based on this type of double-sided support, the pitch support of this invention forms a series compensation relationship with the roll compensation ring.

[0105] The carrier cabin is installed between the pitch supports and houses the multi-mode camera, common reference load frame, fine-tuning frame, and corresponding cable connection structures. The carrier cabin's outer shell can be formed by combining a ball drum, front cover, rear cover, and support plate. Conductive rubber ropes, sealing rings, and anti-loosening fasteners are installed at the outer shell interfaces, giving the carrier cabin waterproof, dustproof, and salt spray resistant capabilities. Multi-mode camera refers to multiple camera payloads used for different marine data acquisition scenarios, including at least visible light, low-light, and infrared cameras. Visible light cameras are suitable for daytime, long-distance, and fog-penetrating scenarios; low-light cameras are suitable for low-light and color night vision scenarios; and infrared cameras are suitable for nighttime thermal feature acquisition and low-visibility scenarios. In this document, the carrier cabin is used to carry optical payloads such as visible light, low-light, and infrared cameras, and is integrated through optical windows, image processing modules, and dedicated structural components. The carrier cabin maintains a consistent load-bearing relationship with the base load.

[0106] The common reference load frame is fixed inside the load cell and adopts either an integrated machined base or a precision-positioned combined bracket. The common reference load frame is equipped with a unified positioning surface, positioning pins, and load clamping components. Visible light imaging, low-light imaging, and infrared imaging are respectively mounted on their corresponding positioning surfaces. Fixed line-of-sight correspondence means that the optical axis direction, imaging center, and angular deviations between the multi-mode cameras remain fixed after assembly calibration and do not change uncertainly with channel switching. To achieve fixed line-of-sight correspondence, a long-distance calibration target can be used during the assembly stage to align the visible light, low-light, and infrared cameras with the same calibration point. The azimuth and pitch deviations of each acquisition channel relative to the unified line-of-sight are recorded and written into the line-of-sight calibration table. The common reference load frame can also be equipped with a thermal isolation plate. The cooling components of the infrared camera are separated from the visible light and low-light cameras by the thermal isolation plate to prevent the heat generated by the infrared camera from affecting the imaging stability of other acquisition channels.

[0107] The fine-tuning frame is housed within the carrier chamber, located between the common reference load frame and the outer shell of the carrier chamber, or between the multi-mode camera mounting end and the common reference load frame. The fine-tuning frame is used for small-angle, rapid, and low-inertia line-of-sight correction. The azimuth mount, roll compensation ring, and pitch support handle large-range and main body attitude compensation; however, these components drive a large structural inertia, which can easily lead to response lag when high-speed, small targets deviate for short periods. The fine-tuning frame can be a two-degree-of-freedom small-angle frame or a structure with flexible hinges and micro-motors. The fine-tuning angle fed back by the fine-tuning frame indicates the current small-angle correction state of the fine-tuning frame to the multi-mode camera's line-of-sight. After the azimuth mount completes horizontal rotation, the roll compensation ring completes reverse compensation, and the pitch support completes vertical pointing, if the preset target in the acquired image still deviates from the center of the field of view, the fine-tuning frame performs angle correction based on the remaining line-of-sight deviation, keeping the preset target within the effective imaging area of ​​the multi-mode camera.

[0108] The aiming control compartment is located inside the azimuth mount or on the fixed side of the azimuth mount. The aiming control compartment can adopt a sealed electronic compartment structure, internally housing an attitude interface board, a trajectory interface board, a turntable drive board, a data processing board, and a power supply protection board. The attitude interface board is used to receive hull attitude data, including roll, pitch, heading angles, and corresponding angular velocities. The trajectory interface board is used to receive pre-pointing parameters, including pre-azimuth, pre-roll, and pre-pitch angles. The turntable drive board outputs azimuth, roll, pitch, and fine-tuning sub-control parameters. The power supply protection board provides overvoltage protection, surge protection, and motor drive power supply isolation. When the aiming control compartment is located inside the azimuth mount, it reduces external cable exposure; when located on the fixed side of the azimuth mount, it facilitates maintenance and heat dissipation. The aiming control compartment is connected to the slip ring, azimuth encoder, roll encoder, roll gyroscope, pitch encoder, and fine-tuning frame via control lines, allowing all feedback data to enter the same control link.

[0109] When the turntable mechanism is working, the pre-pointing parameters are first decomposed into the target angles corresponding to the azimuth mount, roll compensation ring, and pitch support. The pre-azimuth angle is used to determine the target rotation direction of the azimuth mount, the pre-roll angle is used to determine the reverse compensation angle of the roll compensation ring, and the pre-pitch angle is used to determine the target pitch direction of the pitch support. The azimuth encoder provides feedback on the current azimuth angle, the roll encoder provides feedback on the current roll compensation angle, the roll gyroscope provides feedback on the current roll angular velocity, the pitch encoder provides feedback on the current pitch angle, and the fine-tuning frame provides feedback on the current fine-tuning angle. Based on the deviation between the target angle and the feedback angle, azimuth sub-control parameters, roll sub-control parameters, pitch sub-control parameters, and fine-tuning sub-control parameters are generated. Sub-control parameters refer to the control signals used to drive different moving parts of the turntable mechanism. The azimuth sub-control parameter corresponds to the azimuth motor, the roll sub-control parameter corresponds to the roll motor, the pitch sub-control parameter corresponds to the pitch motor, and the fine-tuning sub-control parameter corresponds to the fine-tuning frame.

[0110] After the azimuth control input is output, the azimuth motor drives the azimuth axis system to rotate, which in turn drives the roll compensation ring, pitch support, carrier cabin, and multi-mode camera assembly to rotate in the horizontal plane. After the roll control input is output, the roll motor drives the roll compensation ring to rotate relative to the azimuth mount, providing reverse compensation for the line-of-sight tilt caused by the ship's roll. After the pitch control input is output, the pitch motor drives the pitch axis system to rotate, which in turn drives the carrier cabin to complete the vertical orientation. After the fine-tuning control input is output, the fine-tuning frame makes small-angle corrections to the line-of-sight of the multi-mode camera. These four types of actions are completed collaboratively according to the structural hierarchy: the azimuth mount handles large-scale horizontal tracking, the roll compensation ring handles ship roll disturbances, the pitch support handles target height and direction tracking, and the fine-tuning frame handles residual line-of-sight deviations caused by acquisition channel switching or short-term shifts of high-speed small targets.

[0111] During the acquisition of high-speed small targets at sea, the target, such as an unmanned surface vessel, may move rapidly laterally on the waves, experiencing short-term deviations due to wind and waves. After the acquisition equipment generates a pre-pointing value based on the pre-trajectory, the azimuth mount first rotates towards the future azimuth of the target. The roll compensation ring performs reverse compensation based on the roll angle and roll rate. The pitch support adjusts the bearing chamber according to the future height direction of the target. The fine-tuning frame quickly corrects small angular deviations in the image feedback. If the acquisition channel switches from visible light imaging to low-light imaging or infrared imaging, the fine-tuning frame compensates for the line-of-sight deviation based on the line-of-sight calibration table corresponding to the common reference load frame, ensuring that the acquisition channel still points to the same target after the switch. This working method differs from the simple horizontal rotation and pitch following of ordinary camera turntables. It combines the ship's installation reference, azimuth rotation, roll reverse compensation, pitch pointing, multi-mode camera common reference installation, and line-of-sight fine-tuning into a single turntable mechanism, making it more suitable for scenarios involving large waves, high-speed small targets, and continuous acquisition across different scenes.

[0112] In one possible implementation, the pre-trajectory typically consists of multiple pre-trajectory points arranged in chronological order, with each pre-trajectory point corresponding to a prediction time interval. When a control cycle arrives, the corresponding pre-trajectory point is selected according to the turntable response time of the current control cycle. For high-speed unmanned surface vessels or low-altitude small aircraft, the pre-trajectory points should not be selected too far away; the prediction time interval is usually set to 0.05s to 0.30s. This range mainly considers the image frame period, motor response delay, and target speed changes. Below 0.05s, the lead is not significant, and the turntable still exhibits follow control; above 0.30s, the turning error of high-speed small targets is easily amplified. If the image frame rate of the acquisition device is 25fps to 60fps, and the effective response time of the azimuth and pitch motors is approximately 40ms to 120ms, then a prediction time interval of 0.08s to 0.20s is generally more prudent.

[0113] After reading the pre-trajectory point, the difference between the pre-trajectory point and the installation coordinates is used to obtain the relative position. This relative position represents the spatial vector of the preset target's future position relative to the optical center of the acquisition device. The conversion expression for the relative position is:

[0114] in, Indicates the relative position corresponding to the predicted time; The symbol represents the pre-trajectory point in the pre-trajectory; C represents the installation coordinates of the optical center of the acquisition device in the ship's coordinate system; the subscript t represents the time corresponding to the current control cycle. This represents the prediction time interval. The expression is derived from the translation relationship of spatial vectors. The pre-trajectory point represents the future position of the preset target in the ship's coordinate system, and the installation coordinates represent the position of the optical center of the acquisition equipment. Subtracting the two yields the spatial vector pointing from the acquisition equipment to the future position of the preset target. This spatial vector is then directly used to calculate the azimuth, pitch, and roll compensation angles.

[0115] Ship attitude data is used for attitude compensation in relative position calculations. In wind and waves, ships experience changes in heading, pitch, and roll. The heading angle alters the horizontal pointing of the acquisition equipment relative to the target, the pitch angle causes the target to shift vertically in the frame, and the roll angle causes the line of sight to tilt laterally. If the pre-pointing value is directly generated from the relative position, the instantaneous ship attitude may be mistaken for target movement in high winds and waves, leading to overcompensation in the turntable control. Therefore, before generating the pre-pointing value, the relative position needs to be compensated according to the reverse relationship of the ship attitude.

[0116] Attitude compensation can be achieved using a rotation matrix, and the expression for the compensation position is: in, Indicates the compensation position corresponding to the predicted time; Indicates the relative position corresponding to the predicted time; Indicates the heading angle corresponding to the current control cycle; This indicates the pitch angle corresponding to the current control cycle; Indicates the roll angle corresponding to the current control cycle; This indicates the reverse rotation relationship with respect to the heading angle; This indicates the reverse rotation relationship with respect to the pitch angle; This expression represents the reverse rotation relationship with respect to the roll angle. It corrects the relative position in the opposite direction to the ship's current attitude, essentially pulling the observed coordinates, which sway with the ship, back to a stable attitude, making the compensated position closer to the position the data acquisition equipment should point to in a stable state.

[0117] The effective values ​​for bow angle, pitch angle, and roll angle are typically obtained from inertial measurement units. During routine navigation at sea, the roll angle is usually in the range of -10° to 10°, and the pitch angle is usually in the range of -6° to 6°. When tracking in rough seas, the roll angle can be widened to -25° to 25°, and the pitch angle can be widened to -15° to 15°. This range is mainly determined based on the actual compensation capability and safe movement range of the shipborne electro-optical turntable. When this range is exceeded, although the turntable can still continue to output compensation commands, the carrier compartment may approach the mechanical limit, usually requiring a reduction in tracking speed or the initiation of protection control.

[0118] After the compensated position is determined, the pre-azimuth and pre-pitch angles are calculated based on its components in the ship's coordinate system. The compensated position can be expanded as follows:

[0119] in, This indicates the component of the compensation location along the bow direction; Indicates the component of the compensation position along the starboard side; This represents the vertical component of the compensated position. The three components are directly obtained from the spatial vector after attitude compensation and are used to determine the horizontal rotation of the azimuth mount, the vertical rotation of the pitch support, and the roll correction of the roll compensation ring, respectively.

[0120] The pre-azimuth angle is determined by two components in the horizontal plane, and the expression is: in, This indicates the pre-azimuth angle corresponding to the predicted time. Indicates the component of the compensation position along the starboard side; This represents the component of the compensation position along the bow direction. This expression is derived from the planar polar angle relationship. The projection of the compensation position onto the horizontal plane determines the angle by which the acquisition device should rotate left or right. Quadrant information can be preserved to avoid angle bounce when the preset target is located at the junction of the two sides of the bow.

[0121] The pitch angle is determined by the altitude component and the horizontal distance, and the expression is: in, Indicates the pre-elevation angle corresponding to the predicted time; Indicates the vertical component of the compensation location; and The horizontal distance of the compensation position is determined jointly. This expression is derived from a right triangle relationship. The ratio of the height component to the horizontal distance reflects the vertical angle of the preset target relative to the optical center of the acquisition device. The closer the preset target is to the acquisition device and the more obvious the height difference, the more sensitive the pre-tilt angle changes.

[0122] The pre-roll angle is not determined by the target position itself, but rather by the roll angle and roll rate. The roll compensation loop is positioned between the azimuth mount and the pitch mount, and its function is to counteract the image rotation and lateral slippage of the line of sight caused by the ship's roll. For high-speed, small target acquisition, simply using the current roll angle for compensation will result in a slight lag; therefore, the roll rate needs to be included in the prediction compensation. The expression for the pre-roll angle is:

[0123] in, Indicates the pre-roll angle corresponding to the predicted time; Indicates the roll angle corresponding to the current control cycle; This indicates the roll rate corresponding to the current control cycle; Indicates the prediction time interval; This represents the roll rate coefficient, ranging from 0.5 to 1.0. The roll rate coefficient controls the intensity of roll angular velocity compensation. A value below 0.5 results in insufficient compensation for rapid rolls, and lateral slippage may still occur in the image. A value above 1.0 can easily lead to overcompensation during sudden changes in roll angular velocity. Therefore, during sea trials, it is generally adjusted starting from around 0.7, increasing appropriately when there are significant waves and decreasing appropriately when the sea state is relatively stable. The first term of this expression is used to offset the current roll angle, and the second term is used to estimate the roll change within the prediction time interval, causing the roll compensation loop to act in advance before the next control cycle. The pre-pointing parameter is formed by combining the pre-azimuth angle, pre-pitch angle, and pre-roll angle, and its expression is:

[0124] in, This indicates the pre-pointing quantity corresponding to the predicted time. Indicates the pre-azimuth angle; Indicates the pre-tilt / pitch angle; This indicates the pre-roll angle. The pre-pointing value essentially converts the pre-trajectory point into an angular target that the turntable mechanism can execute, corresponding to the target angles of the azimuth mount, pitch support, and roll compensation ring, respectively.

[0125] After the pre-aiming signal is sent to the pre-aiming control chamber, it is compared with feedback data from the azimuth encoder, pitch encoder, roll encoder, and roll gyroscope. The azimuth encoder provides feedback on the current azimuth angle of the azimuth mount, the pitch encoder provides feedback on the current pitch angle of the pitch support, the roll encoder provides feedback on the current roll compensation angle of the roll compensation ring, and the roll gyroscope provides feedback on the roll angular velocity near the roll compensation ring. The encoders provide positional feedback, while the gyroscopes provide angular velocity feedback. Using these two types of feedback together allows the turntable control to know both the current angular deviation and whether the angular deviation is increasing or decreasing. The azimuth deviation, pitch deviation, and roll deviation can be expressed as follows:

[0126] in, This indicates the azimuth deviation corresponding to the current control cycle; This indicates the pitch deviation corresponding to the current control cycle; This indicates the roll deviation corresponding to the current control cycle; Indicates the azimuth angle fed back by the azimuth encoder; This indicates the pitch angle fed back by the pitch encoder; This indicates the roll compensation angle provided by the roll encoder feedback; This indicates that the azimuth difference is converted to the shortest rotation interval. The azimuth needs to be processed for the shortest rotation because the azimuth seat can rotate continuously. If the angle is not normalized, the change of the pre-azimuth angle from 179° to -179° can easily be misjudged as a large angle rotation of 358°.

[0127] In practical control, the turntable control quantity should not be directly output based solely on the angle deviation. During high-speed, small target acquisition, the azimuth mount and pitch support bear the large inertia of structural motion, while the roll compensation loop handles rapid reverse compensation. Since their response speeds and loads differ, the variation range of the control quantities needs to be limited separately. The azimuth, pitch, and roll sub-control quantities can be generated using the following formula:

[0128] in, This indicates the azimuth sub-control quantity corresponding to the current control cycle; This represents the pitch control variable corresponding to the current control cycle; This indicates the horizontal rocker control quantity corresponding to the current control cycle; , , These represent the angular scaling factors for azimuth, pitch, and roll, respectively. , These represent the deviation coefficients for azimuth and elevation, respectively. Indicates the roll rate coefficient; Indicates the control cycle; , , These represent the upper limits of control for azimuth, pitch, and roll, respectively. This indicates amplitude limiting. The proportional term provides the basic driving force based on the angular deviation, the deviation variation term suppresses overshoot, and the roll rate term allows the roll compensation loop to respond to hull roll changes in advance. Amplitude limiting prevents excessive instantaneous motor drive, avoiding impacts to the load cell or distortion of encoder feedback.

[0129] The above coefficients can be calibrated according to the load weight and motor response capability during overall machine commissioning. The azimuth proportionality coefficient is typically set to 0.6 to 1.5, the pitch proportionality coefficient to 0.5 to 1.2, and the roll proportionality coefficient to 0.8 to 2.0. With larger loads in the azimuth and pitch directions, excessively high proportionality coefficients will cause the load-bearing compartment to sway, while excessively low coefficients will cause lag in tracking high-speed small targets. The roll compensation loop inertia is relatively small, allowing for higher response coefficients. The azimuth deviation variation coefficient and pitch deviation variation coefficient are typically set to 0.02 to 0.20 to reduce overshoot; they should not be too high, otherwise they will suppress normal tracking response. The roll rate coefficient is typically set to 0.05 to 0.30, with the value determined based on the roll gyroscope noise and the rate of roll change; higher values ​​are used when there are significant waves, and lower values ​​are used when the sea state is stable.

[0130] The control cycle is typically set between 10ms and 40ms. Below 10ms, the requirements for ordinary shipboard processing hardware and encoder feedback links are higher, and noise is introduced more frequently; above 40ms, turntable control lag is likely to occur in scenarios involving high-speed, small targets such as unmanned surface vessels and low-altitude small aircraft. In engineering debugging, 20ms is commonly used as the initial control cycle, and then fine-tuned based on the response status of the azimuth, pitch, and roll motors. The turntable control quantity is formed by combining multiple sub-control quantities, expressed as:

[0131] in, This indicates the turntable control quantity corresponding to the current control cycle; Indicates the azimuth control quantity; Indicates pitch control quantity; Indicates the horizontal control quantity; This represents the fine-tuning control values. The azimuth control value drives the azimuth mount, the pitch control value drives the pitch mount, the roll control value drives the roll compensation ring, and the fine-tuning control value drives the fine-tuning frame. The fine-tuning control values ​​are typically determined by the unallocated small-angle deviations of the azimuth mount and pitch mount, the residual line-of-sight tilt after roll compensation, and the line-of-sight calibration deviations of multi-mode cameras.

[0132] In one possible implementation, within a control cycle, the pre-azimuth angle, pre-roll angle, and pre-pitch angle are first read. Simultaneously, the azimuth angle fed back by the azimuth encoder, the roll compensation angle fed back by the roll encoder, the roll angular velocity fed back by the roll gyroscope, the pitch angle fed back by the pitch encoder, and the fine-tuning angle fed back by the fine-tuning frame are also read. The pre-azimuth angle represents the horizontal pointing angle that the azimuth mount needs to reach in advance during the current control cycle; the azimuth angle represents the current actual rotation angle of the azimuth axis system. The pre-roll angle represents the reverse compensation angle that the roll compensation loop needs to form; the roll compensation angle represents the current actual compensation angle of the roll compensation loop; the roll angular velocity represents the rate of change of the ship's roll. The pre-pitch angle represents the vertical pointing angle that the pitch support needs to reach in advance; the pitch angle represents the current actual rotation angle of the pitch axis system. The fine-tuning angle represents the small-angle line-of-sight correction amount that the fine-tuning frame has already formed. The above feedback is generally updated with a control cycle of 10ms to 40ms. This range mainly takes into account the encoder feedback speed, motor response time, and the computing load of the ship's controller. If the cycle is too long, high-speed small targets are likely to deviate from the field of view before the next frame is acquired. If the cycle is too short, the feedback noise is likely to be amplified.

[0133] Azimuth deviation is calculated based on the shortest rotation direction. Since the azimuth seat has a continuous rotation structure, directly subtracting the pre-azimuth angle from the azimuth angle can easily lead to large-angle misjudgments at the boundary between positive and negative angles. For example, a change from 179° to -179° actually only requires a small-angle rotation. Therefore, the azimuth deviation is obtained through angle normalization, and its calculation formula is:

[0134] in, Indicates the pre-azimuth angle. This indicates the azimuth angle fed back by the azimuth encoder. This indicates the azimuth deviation, and `wrap` means normalizing the angle difference to the shortest turning interval. Essentially, this calculation method allows the azimuth to complete its rotation along the shortest path, avoiding large reverse movements due to angle crossings.

[0135] After obtaining the azimuth deviation, it is compared with the azimuth fine-tuning threshold. The azimuth fine-tuning threshold is used to define the action boundaries of the azimuth seat and the fine-tuning frame. It is generally set to 5% to 20% of the horizontal field of view of the multi-mode camera, and is usually 0.2° to 1.5° in long-distance, high-speed, small target acquisition. This range is set considering two points: if the value is too small, the azimuth seat will frequently respond to very small target jitters, which can easily cause the carrier to swing back and forth; if the value is too large, the fine-tuning frame will have to bear too much azimuth correction, which may exceed the small angle compensation capability. During debugging, you can start with a setting of 0.5°. If the azimuth seat moves too frequently, increase it appropriately; if the target is prone to deviating from the center of the image, decrease it appropriately.

[0136] The portion of the azimuth deviation exceeding the azimuth fine-tuning threshold is compensated by the azimuth seat, while the portion below the threshold is handled by the fine-tuning frame. The azimuth seat compensation angle can be determined by the following formula:

[0137] in, Indicates the azimuth compensation angle. Indicates azimuth deviation. denoted by , sgn represents the azimuth fine-tuning threshold, sgn represents the sign function, and max represents taking the larger value. The processing logic of this formula is that when the azimuth deviation is small, it does not drive the azimuth seat to perform a large-range movement; when the azimuth deviation exceeds the azimuth fine-tuning threshold, the excess angle is then handed over to the azimuth seat to complete the main body rotation.

[0138] The azimuth control quantity is determined jointly by the azimuth seat compensation angle and the change in the azimuth seat compensation angle. The azimuth seat bears the roll compensation ring, pitch support, and bearing cabin, and has a large inertia. Therefore, a large drive quantity cannot be directly given based on the azimuth deviation; otherwise, oscillation will occur. The azimuth control quantity can be formed by the following formula:

[0139] in, Indicates the azimuth control quantity. This represents the azimuth ratio coefficient, with a value ranging from 0.6 to 1.5. This represents the azimuth variation coefficient, with a value ranging from 0.02 to 0.20. This indicates the azimuth compensation angle of the previous control cycle. Indicates the control period. This indicates the upper limit of the azimuth control quantity, and `sat` indicates amplitude limiting. The azimuth ratio coefficient determines the force with which the azimuth seat follows the azimuth deviation; a value that is too small will cause azimuth rotation lag, while a value that is too large can easily cause overshoot. The azimuth variation coefficient is used to suppress the azimuth seat from rapidly crossing the target direction; setting it too high will suppress normal tracking response. Amplitude limiting is used to protect the azimuth motor and azimuth axis system, avoiding instantaneous impacts during strong winds, waves, and rapid lateral movement of the target.

[0140] The roll deviation is obtained by directly subtracting the pre-roll angle and the roll compensation angle. The goal of the roll compensation ring is to counteract the line-of-sight tilt caused by the ship's roll; therefore, the pre-roll angle is usually opposite to the direction of the ship's roll. The formula for calculating the roll deviation is:

[0141] in, Indicates the pre-roll angle. This indicates the roll compensation angle provided by the roll encoder. This indicates the roll deviation. This formula reflects the angle difference that the roll compensation ring still needs to compensate for. The larger the roll deviation, the more obvious the current insufficiency of reverse compensation.

[0142] In addition to using the roll deviation, the roll rate must also be considered in the roll control parameters. The hull rolls rapidly in cross waves; if only roll deviation is used for compensation, the roll compensation loop will always lag by one control cycle. By incorporating the roll rate, the direction of the roll change can be predicted in advance, allowing the roll compensation loop to pre-adjust or reduce the compensation amount. The roll control parameters can be formed using the following formula:

[0143] in, This indicates the amount of horizontal control. This represents the roll ratio factor, with values ​​ranging from 0.8 to 2.0. This represents the roll rate coefficient, with values ​​ranging from 0.05 to 0.30. This indicates the roll angular velocity fed back by the roll gyroscope. This indicates the upper limit of the roll control. The roll ratio coefficient can be higher than the bearing ratio coefficient because the roll compensation ring has a relatively small load and needs to quickly counteract the ship's roll. The roll angular velocity coefficient is used to respond to roll changes in advance. A higher value can be used when there are obvious cross waves, and a lower value should be used when the sea state is relatively stable to avoid amplifying the noise of the roll gyroscope.

[0144] Pitch deviation is obtained by subtracting the pre-pitch angle from the pitch angle. The pitch support rotates the support tank up and down to compensate for vertical offsets caused by changes in the preset target height, hull pitch, and wave surface fluctuations. The formula for calculating pitch deviation is:

[0145] in, Indicates the pitch angle. This indicates the pitch angle fed back by the pitch encoder. This indicates the pitch deviation. The formula represents the difference between the current actual angle of the pitch support and the target angle, used to determine the rotation direction and amplitude of the pitch motor.

[0146] After comparing the pitch deviation with the pitch fine-tuning threshold, the portion exceeding the threshold is compensated by the pitch support, while the remaining portion is handled by the fine-tuning frame. The pitch fine-tuning threshold is generally set to 5% to 20% of the vertical field of view of the multi-mode camera, with a commonly used range of 0.15° to 1.0°. Since the pitch direction directly affects the vertical retention of sea surface targets in the image, the threshold should not be too large; however, the pitch support also supports the payload bay, and a threshold that is too small will cause frequent actuation of the pitch motor. In high-speed unmanned surface vessel (USV) data acquisition scenarios, the threshold can be adjusted starting from 0.3°, while it can be appropriately reduced in low-altitude small aircraft scenarios.

[0147] The pitch support compensation angle can be determined by the following formula: in, Indicates the pitch support compensation angle. Indicates pitch deviation. This indicates the pitch fine-tuning threshold. This formula assigns large-angle deviations in the pitch direction to the pitch support, making the pitch support responsible for bearing the orientation of the main body of the cabin, while small-angle residuals are quickly corrected by the fine-tuning frame.

[0148] The pitch control quantity can be formed by the following formula: in, This represents the pitch control variable. This represents the pitch ratio factor, with a value ranging from 0.5 to 1.2. This represents the pitch variation factor, with values ​​ranging from 0.02 to 0.20. This indicates the pitch support compensation angle in the previous control cycle. This indicates the upper limit of the pitch control. The pitch ratio factor is usually slightly lower than the roll ratio factor because the load inertia of the carrier in the pitch direction is larger, and too high a value would cause the vertical direction to swing back; the pitch variation factor is used to suppress pitch support overshoot, so that low-altitude small aircraft or fast-drifting targets on the sea surface remain stable in the vertical direction of the image.

[0149] The fine-tuning control values ​​are determined after the azimuth, yaw, and pitch control values ​​are established. The fine-tuning frame is only responsible for small-angle line-of-sight correction and does not handle large-angle movements of the azimuth mount and pitch support. First, the remaining portion of the azimuth deviation not compensated by the azimuth mount is calculated, then the remaining portion of the pitch deviation not compensated by the pitch support is calculated, and the remaining line-of-sight tilt after reverse compensation by the yaw compensation loop is read. If the current acquisition channel is switched, the line-of-sight deviation of the current acquisition channel also needs to be read from the line-of-sight calibration table corresponding to the multi-mode camera.

[0150] The azimuth residual error and pitch residual error can be expressed as: in, This indicates the remaining azimuth deviation. Indicates the remaining pitch deviation. Indicates azimuth deviation. Indicates pitch deviation. Indicates the azimuth compensation angle. This indicates the pitch support compensation angle. This calculation method allows the fine-tuning frame to only accept small angular deviations that are not processed by the azimuth and pitch supports, avoiding repeated compensation for the same angular deviation in multiple places.

[0151] The line-of-sight calibration table for multi-mode cameras is established during assembly and debugging. In actual products, although visible light, low-light, and infrared cameras are mounted on a common reference load cell, slight installation deviations still exist in the optical axes of each acquisition channel. By acquiring data from the same distant target in the calibration field, the azimuth, elevation, and line-of-sight tilt calibration deviations between different acquisition channels can be obtained and written into the line-of-sight calibration table. The commonly used line-of-sight calibration deviation range is -0.3° to 0.3°, which mainly comes from lens mounting tolerances, common reference load cell machining errors, and thermal deformation errors. When the deviation exceeds this range, it is usually necessary to readjust the multi-mode camera mounting position rather than simply relying on fine-tuning frame compensation.

[0152] Fine-tuning deviations can be formed using the following formula: in, Indicates fine-tuning deviation. This indicates the remaining azimuth deviation. Indicates the remaining pitch deviation. This indicates the remaining tilt of the line of sight after the reverse compensation of the roll compensation ring. This represents the line-of-sight calibration deviation corresponding to the current acquisition channel. This formula combines the residual deviation after the subject points and the inherent deviation of the acquisition channel to obtain the target deviation that the fine-tuning frame needs to correct.

[0153] The fine-tuning control value can be formed by the following formula: in, This indicates the fine-tuning control value. This represents the fine-tuning ratio, with a value ranging from 0.8 to 2.5. This indicates the magnitude of the fine-tuning deviation. This indicates the upper limit of the fine-tuning control. The value of the fine-tuning ratio coefficient can be higher than that of the azimuth ratio coefficient and the pitch ratio coefficient because the fine-tuning frame only drives the light-load, small-angle mechanism, resulting in a faster response speed; however, if the value is too high, small jitters may occur when switching acquisition channels or when there is a lot of noise in the infrared image. Therefore, during sea trials, the value is usually adjusted from 1.2 to 1.6.

[0154] After the azimuth mount, roll compensation ring, pitch support, and fine-tuning frame complete their actions, the feedback results from the azimuth encoder, roll encoder, roll gyroscope, pitch encoder, and fine-tuning frame are read. These feedback results are compared with the pre-pointing parameters to obtain the residual deviation. The residual deviation is used to determine the actual compensation effect for the current control cycle and also to update the azimuth, roll, pitch, and fine-tuning sub-control parameters for the next control cycle.

[0155] The residual deviation can be expressed by the following formula: in, Indicates residual deviation. This indicates the azimuth angle fed back by the azimuth encoder after the action is completed. This indicates the pitch angle fed back by the pitch encoder after the action is completed. This indicates the roll compensation angle provided by the roll encoder after the action is completed. This represents the residual amount of fine-tuning feedback after the action is completed. This formula compares the actual feedback after the action is completed with the pre-pointing amount to determine whether each execution part has reached the expected point of reference.

[0156] The sub-control quantity for the next control cycle can be adjusted by adding residual deviation correction to the previous control quantity, as expressed in the following expression: in, This represents the set of sub-control variables corresponding to the next control cycle. This represents the set of sub-control variables corresponding to the current control cycle. Indicates residual deviation. This represents the residual correction factor, with a value ranging from 0.1 to 0.5. The residual correction factor should not be too high, as the residual deviation includes encoder noise, roll gyroscope noise, and structural elastic hysteresis; an excessively high value will cause control oscillations. Conversely, a value that is too low will result in slower convergence of continuous residual deviations. During engineering debugging, a value of 0.2 can be initially used, and gradually increased if the residual deviation fails to converge for three consecutive control cycles.

[0157] In one possible implementation, after the turntable control quantities are generated, they are first written into the turntable mechanism of the acquisition equipment according to the control cycle. The turntable control quantities include azimuth sub-control quantities, roll sub-control quantities, pitch sub-control quantities, and fine-tuning sub-control quantities. The azimuth sub-control quantity is used to control the azimuth mount to complete horizontal rotation; the roll sub-control quantity is used to control the roll compensation ring to counteract the line-of-sight tilt caused by the ship's roll; the pitch sub-control quantity is used to control the pitch support to drive the carrier cabin to rotate up and down; and the fine-tuning control quantity is used to control the fine-tuning frame to make small-angle corrections to the line-of-sight of the multi-mode camera. Here, the line-of-sight refers to the imaging center line of the multi-mode camera. Whether the line-of-sight is stable directly affects whether the preset target can be kept within the effective area of ​​the acquired image.

[0158] In actual operation, the azimuth control signal first acts on the azimuth motor, which drives the azimuth axis system to rotate, causing the azimuth seat to rotate toward the preset azimuth direction of the target. The azimuth axis system is the slewing bearing structure inside the azimuth seat, responsible for supporting the rotation of the upper roll compensation ring, pitch support, and the entire bearing cabin. The azimuth encoder synchronously feeds back the real-time azimuth angle of the azimuth axis system. If the fed-back azimuth angle does not reach the position required by the turntable control signal, the azimuth control signal is continuously corrected; if the azimuth angle enters the allowable error range, the azimuth seat switches from rapid rotation to low-speed following. This process avoids frequent and large movements of the azimuth seat due to short-term left-right swaying of a high-speed, small target.

[0159] The roll control input acts on the roll motor, which in turn drives the roll compensation ring for reverse compensation. Located between the azimuth mount and the pitch support, the roll compensation ring's function is not to change the azimuth of the preset target, but rather to keep the line of sight of the load cell as stable as possible when the ship tilts to the left or right. The roll encoder provides feedback on the current compensation angle of the roll compensation ring, while the roll gyroscope provides feedback on the roll rate. The roll rate indicates how quickly the ship's roll changes. When the roll rate is high, the roll compensation ring increases the compensation amplitude in advance; when the roll rate decreases and approaches stability, the roll compensation ring decreases the compensation amplitude to prevent the image from wobbling in the opposite direction.

[0160] The pitch control variable acts on the pitch motor, which drives the pitch axis system to rotate, causing the carrier to pitch up or down according to the pre-set pitch direction. The pitch axis system is the rotating structure that supports the carrier on both sides of the pitch bracket. Multiple cameras are fixed inside the carrier, so the movement of the pitch axis system directly changes the vertical orientation of the multiple cameras. The pitch encoder provides real-time pitch angle feedback from the pitch axis system. If the preset target is affected by wave undulations, ship pitch, or its own maneuvering and shifts vertically in the image, the pitch control variable continuously corrects the vertical angle of the carrier to bring the preset target back to the center of the captured image.

[0161] After the azimuth mount, roll compensation ring, and pitch support complete the main pointing, the fine-tuning control is applied to the fine-tuning frame. The fine-tuning frame, located within the carrier chamber, is used for small-angle line-of-sight correction of multi-mode cameras. The fine-tuning frame does not handle large-angle rotation; its main function is to address small-angle deviations not compensated by the azimuth mount and pitch support, as well as minor line-of-sight differences between visible light, low-light, and infrared cameras. This way, when switching acquisition channels, there is no need to re-search a large area for the preset target; the fine-tuning frame only needs to make fine adjustments based on the line-of-sight calibration relationship to ensure that the switched acquisition channel continues to align with the same preset target.

[0162] After the turntable mechanism completes its pointing, the actual acquisition channel is determined based on current marine environmental data, preset target status, and acquisition channel identification. Multi-mode imaging includes visible light imaging, low-light imaging, and infrared imaging. Visible light imaging is suitable for daytime, high-visibility scenarios where target details need to be preserved; low-light imaging is suitable for twilight, low-light nighttime scenarios where the target still retains some contour contrast; infrared imaging is suitable for nighttime, rain / fog, scenarios where the target's thermal characteristics are obvious, or scenarios with insufficient visible light contrast. Acquisition channels are not switched in isolation, but rather the pre-trajectory and turntable control values ​​from the previous control cycle are used to maintain target identity and line-of-sight continuity.

[0163] After determining the acquisition channel, the imaging parameters for the selected channel are set. For visible light imaging, the main settings are focal length, focus state, exposure time, and fog penetration mode; for low-light imaging, the main settings are gain, brightness equalization, and low-light enhancement mode; for infrared imaging, the main settings are focal length, focus state, polarity mode, and contrast. Focal length controls the size of the target image in the frame, focus state ensures sharp target edges, exposure time and gain adjust image brightness, and polarity mode selects between incandescent and abbreviated display in the infrared image. These parameters are adjusted according to target distance, target speed, and changes in the marine environment to prevent blurry images of high-speed, small targets at long distances or in low light conditions.

[0164] At the start of the acquisition, the multi-mode camera acquires images according to the current orientation of the turntable mechanism, recording the image frame along with the acquisition time, acquisition channel identifier, azimuth angle, pitch angle, roll compensation angle, and ship attitude data. The acquisition time is used to mark the moment corresponding to the image frame, the acquisition channel identifier is used to indicate whether the image frame originated from visible light imaging, low-light imaging, or infrared imaging, and the azimuth angle, pitch angle, and roll compensation angle are used to record the actual attitude of the acquisition equipment at that time. Subsequent target verification, trajectory playback, or evidence chain organization can clearly identify the turntable state and ship state corresponding to each image frame.

[0165] During continuous acquisition, the position of a preset target in the acquired image is detected. If the preset target is located in the central area of ​​the acquired image, it indicates that the turntable control is basically matched with the actual movement of the preset target, and the acquisition device continues to operate with the current acquisition channel and the current turntable control. If the preset target shifts towards the edge of the image, the actual deviation of the preset target relative to the center of the image is extracted. The actual deviation represents the direction and magnitude of the preset target's offset in the image coordinates, reflecting the residual difference between the current turntable control and the actual target position.

[0166] When the actual deviation is small, the fine-tuning frame prioritizes small-angle corrections. This reduces frequent movements of the azimuth and pitch supports, ensuring image stability. When the actual deviation continues to increase, or the preset target approaches the edge of the captured image, the azimuth and pitch supports are then used for main body adjustments. If the actual deviation is accompanied by image tilt, the roll compensation loop performs reverse compensation simultaneously. This tiered action method can balance wide-range tracking and small-angle image stabilization, making it particularly suitable for high-speed, small targets such as unmanned surface vessels, low-altitude small aircraft, and rapidly drifting targets on the sea surface.

[0167] When a preset target is briefly obstructed by waves, rain, sea fog, or ship hulls, the target is not immediately declared lost. Instead, the current turntable control is maintained for a short holding window. This holding window can be determined based on the target speed and acquisition channel, and is typically set to 0.2s to 1.0s in engineering practice. This range is chosen because high-speed, small targets are usually obscured by waves for a short time. If the holding window is less than 0.2s, frequent re-searching is likely; if it is greater than 1.0s, the turntable may continue pointing in the wrong direction, affecting re-acquisition efficiency. Within the short holding window, the acquisition device continues acquisition in the direction corresponding to the pre-trajectory. Once the preset target reappears in the expected area, acquisition resumes.

[0168] When switching acquisition channels, the current main orientation of the azimuth mount, roll compensation ring, and pitch support is maintained first. Then, the fine-tuning frame compensates according to the line-of-sight calibration relationship of the multi-mode cameras. The line-of-sight calibration relationship is obtained during the assembly and debugging phase by calibrating the visible light camera, low-light camera, and infrared camera using the same target. It is used to represent the fixed deviation of the line-of-sight between different acquisition channels. In this way, when switching from visible light camera to low-light camera, or from low-light camera to infrared camera, the preset target will not suddenly shift out of the frame due to differences in the physical installation of the acquisition channels.

[0169] Once acquired, the image frames are stored in the image buffer and associated with the target record. The target record includes the preset target number, acquisition time, acquisition channel identifier, target image position, target confidence level, turntable status, and ship attitude data. The target confidence level indicates the reliability of the preset target recognition result in the current image frame, and is typically determined by the target contour sharpness, target thermal features, target position continuity, and image noise level. If the target confidence level is high, the current image frame can be directly written into the valid acquisition sequence; if the target confidence level is low, it is retained as an image frame to be reviewed, awaiting continuity confirmation from subsequent image frames.

[0170] In one possible implementation, after the acquisition device enters continuous tracking, it first reads the aligned marine environmental data and the status of the preset target within the current control cycle. The marine environmental data mainly includes visibility, ambient light, wind speed, wind direction, wave height, wave direction, wave period, rain / fog intensity, and sea surface reflectivity. Visibility is used to determine whether the outline of a distant target is easily weakened by air moisture; ambient light is used to determine whether the visible light camera has normal imaging conditions; rain / fog intensity is used to determine whether the visible light image will show significant scattering and contrast reduction; and sea surface reflectivity is used to determine whether the preset target is easily interfered with by bright spots from waves or solar reflection. Wind speed, wind direction, wave height, wave direction, and wave period are mainly used to determine the short-term drift risk of the preset target in the image, and also to estimate whether it is necessary to increase exposure, reduce gain, or switch to infrared imaging when switching acquisition channels.

[0171] The status of the preset target is read from the target record, which generally includes target distance, target speed, target size, target confidence level, target image position, target outline sharpness, target thermal features, target channel continuity, and target occlusion status. Target distance indicates the distance of the preset target relative to the acquisition device; target speed indicates the speed of movement of the preset target during continuous acquisition; target size indicates the area occupied by the preset target in the image; target confidence level indicates the reliability of the current recognition result; target image position indicates the position of the preset target relative to the center of the image; target outline sharpness indicates whether edges and details are discernible; target thermal features indicate the thermal difference between the target and the sea background in the infrared image; target channel continuity indicates whether the preset target maintains the same target number in consecutive acquisition channels; target occlusion status indicates whether the preset target is temporarily occluded by waves, rain, fog, ship structures, or other targets.

[0172] When determining the acquisition channel, a simple switch between day and night is not used. Instead, a comprehensive judgment is made based on marine environmental data and the status of the preset target. Visible light imaging is suitable for scenarios with good visibility, sufficient ambient light, clear target outlines, long target distances, and the need to preserve details such as the ship's shape, heading, attitude, and deck structure. In engineering practice, visible light imaging is prioritized when ambient light is above approximately 50 lx, visibility is greater than approximately 5 km, and the target outline is relatively clear. The reason for this range is that visible light imaging can provide the most complete images under good illumination and visibility conditions, facilitating subsequent identification of the preset target's shape, direction of movement, and relative position. If visibility continues to decrease, even if the visible light imaging has zoom capabilities, the image contrast will significantly decrease.

[0173] When sea fog, rain, or water vapor reduces the contrast of visible light images, but the target can still form a stable outline in the visible light image, the visible light camera is still used as the acquisition channel, and fog-penetrating imaging, exposure adjustment, and focus correction are enabled. Fog-penetrating imaging refers to using the near-infrared fog-penetrating band to enhance the target outline, allowing the target to still be acquired in light or moderate fog conditions. This processing is suitable for scenarios with visibility of approximately 1km to 5km, ambient illumination that still meets the requirements for visible light imaging, and where the target edges have not completely disappeared. If the system directly switches to infrared imaging, the target color, hull markings, and appearance details may be lost. Therefore, when the target details are still valuable to preserve, the visible light camera remains the priority acquisition channel.

[0174] Low-light imaging is suitable for target acquisition in low-light, shadowy, or non-active lighting conditions at dusk, night, or in low-light environments. When the ambient illuminance is below approximately 50 lx but above approximately 0.001 lx, and the target still retains a discernible outline or weak reflective features, the acquisition channel switches to low-light imaging. Low-light imaging preserves target color and outline by improving imaging capabilities in low light, making it suitable for continuous acquisition without using searchlights or revealing the ship's position. For low-altitude small aircraft or distant unmanned surface vessels, if the visible light camera image gradually darkens but the target outline is not completely submerged in the background, low-light imaging can smoothly continue acquisition.

[0175] Infrared imaging is suitable for scenarios such as nighttime, low light, significant rain or fog obscuring the target, prominent target thermal features, or insufficient contrast between visible and low-light images. When the ambient light is below approximately 0.001 lx, or when the target outline is unclear but its thermal features are obvious, the acquisition channel switches to infrared imaging. Infrared imaging does not rely on visible light reflection but utilizes the thermal difference between the preset target and the sea surface background for imaging. It is suitable for the detection and continuous confirmation of unmanned surface vessels, small boats with significant engine waste heat, low-altitude small aircraft, and fast-drifting targets on the sea surface at night. If there are many hot spots on the sea surface, it is necessary to combine the target speed, target trajectory continuity, and target image position for judgment to avoid misjudging wave surface thermal noise as the preset target.

[0176] Before switching acquisition channels, check the target confidence level and target image position of the preset target in the current acquisition channel. If the target confidence level only decreases briefly, and the target image position is still close to the center of the image, the acquisition device will maintain the current acquisition channel for a short observation period without immediately switching. The short observation time is usually set to 0.2s to 0.8s. The reason for this range is that image shifts caused by waves, raindrops passing by windows, and ship rolling are often very short-lived, and switching acquisition channels too early can easily cause discontinuities in target numbering; if the short observation time is too long, it may miss the effective sequence opportunity in low light or rain / fog conditions.

[0177] If the target confidence level continuously decreases, and the target outline sharpness, target thermal characteristics, or ambient illumination already meet the activation conditions of another acquisition channel, then acquisition channel switching begins. During switching, the azimuth mount, roll compensation ring, and pitch support maintain the current subject orientation, and the fine-tuning frame makes small-angle corrections based on the line-of-sight calibration relationship of the multi-mode cameras. The line-of-sight calibration relationship is established during the assembly and debugging phase and is used to record the fixed line-of-sight deviation between visible light imaging, low-light imaging, and infrared imaging. After this processing, when the acquisition channel switches, the preset target remains in a position close to the center of the image, eliminating the need for a large-scale re-search.

[0178] After determining the acquisition channel, set the imaging parameters for the selected channel. For visible light imaging, primarily adjust the focal length, focus position, exposure time, shutter speed, and fog penetration mode. Increase the focal length when the target is far away, shorten the exposure time when the target is moving quickly, enable fog penetration mode when sea fog intensifies, and perform focus correction when the target outline softens. For low-light imaging, primarily adjust gain, brightness equalization, noise suppression, and sharpness enhancement. Gain is used to enhance the brightness of low-light images, but excessive gain can introduce noise; therefore, when the target is moving quickly, gain cannot be simply increased, and image motion blur must also be considered. For infrared imaging, primarily adjust the focal length, focus position, polarity mode, contrast, and brightness. Polarity modes include incandescent and black-hot modes. Incandescent modes are suitable for scenes where the target's thermal characteristics are higher than the background, while black-hot modes are suitable for enhancing target boundaries when the background is bright in some areas of the sea.

[0179] After completing the imaging parameter settings, the acquisition device outputs image frames through the acquisition channel and writes the image frames, along with the acquisition time, acquisition channel identifier, target number, azimuth angle, pitch angle, roll compensation angle, target image position, and target confidence level, into the target record. The acquisition time ensures that different image frames can be played back in chronological order; the acquisition channel identifier indicates whether the image frame originated from visible light imaging, low-light imaging, or infrared imaging; the target number maintains consistency in the identity of the same preset target across different acquisition channels; and the azimuth, pitch, and roll compensation angles record the actual pointing state of the turntable mechanism during acquisition.

[0180] During the acquisition process, the system continuously checks whether the preset target is within the effective area of ​​the image. The effective area is generally set to 60% to 80% of the area surrounding the center of the image. This range is set because when the preset target is in the central area, there is still sufficient compensation margin for subsequent fine-tuning; if the preset target is always at the edge of the image, even if the target is still visible in the current image, it is easy to lose it in the next control cycle. If the preset target deviates from the effective area of ​​the image, the fine-tuning frame is used first for small-angle correction; if the fine-tuning frame still cannot return to the effective area after compensation, the azimuth and pitch supports are used to readjust the subject's orientation.

[0181] This embodiment also discloses a ship-based all-weather, multi-scenario data collection device, referring to... Figure 3The device includes an acquisition module 301, a processing module 302, and an output module 303. It is used to execute any of the above-described all-weather, multi-scenario data acquisition methods for ships, wherein: The acquisition module 301 is used to acquire marine environmental data, ship attitude data, and historical trajectory data of preset targets. Processing module 302 is used to generate wind and wave data based on marine environmental data. The wind and wave data is used to characterize the impact of wind and waves on the swaying of the ship and the motion of the preset target. Processing module 302 is used to process historical trajectory data to form the motion trend trajectory of a preset target; The processing module 302 is used to combine wind and wave data with the motion trend track to correct the simulated position of the preset target and generate the preset trajectory of the preset target. The processing module 302 is used to convert the pre-trajectory into a pre-pointing quantity corresponding to the acquisition device, and combine it with the ship attitude data to perform attitude compensation, thereby forming the turntable control quantity of the acquisition device. Output module 303 is used to control the acquisition device to acquire preset targets based on the turntable control quantity.

[0182] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0183] This embodiment also discloses an electronic device, as shown in the reference. Figure 4 The electronic device may include: at least one processor 401, at least one communication bus 402, user interface 403, network interface 404, and at least one memory 405.

[0184] The communication bus 402 is used to enable communication between these components.

[0185] The user interface 403 may include a display screen and a camera. Optionally, the user interface 403 may also include a standard wired interface and a wireless interface.

[0186] The network interface 404 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0187] The processor 401 may include one or more processing cores. The processor 401 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 405, and by calling data stored in memory 405. Optionally, the processor 401 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 401 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip without being integrated into the processor 401.

[0188] The memory 405 may include random access memory (RAM) or read-only memory. Optionally, the memory may include a non-transitory computer-readable storage medium. The memory 405 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 405 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch functionality, sound playback functionality, image playback functionality, etc.), instructions for implementing the various method embodiments described above, etc.; the data storage area may store data involved in the various method embodiments described above, etc. Optionally, the memory 405 may also be at least one storage device located remotely from the aforementioned processor 401. As a computer storage medium, the memory 405 may include an operating system, a network communication module, a user interface 403 module, and an application program for a ship all-weather multi-scenario data acquisition method.

[0189] In the electronic device, the user interface 403 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 401 can be used to call an application program stored in the memory 405 that is a method for collecting data in all weather and multiple scenarios on a ship. When executed by one or more processors 401, the electronic device executes one or more methods as described in the above embodiments.

[0190] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, as some steps can be performed in other orders or simultaneously according to the present invention. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0191] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0192] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between apparatuses or units may be electrical or other forms.

[0193] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0194] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0195] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 405 and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned memory 405 includes various media capable of storing program code, such as a USB flash drive, external hard drive, magnetic disk, or optical disk.

[0196] The present invention also discloses a computer-readable storage medium storing instructions. When executed by one or more processors 401, these instructions cause an electronic device to perform one or more methods as described in the above embodiments.

[0197] The above are merely exemplary embodiments of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of other embodiments of this disclosure upon considering the specification and the disclosure of practical truths. This invention is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for collecting data from ships in all weather conditions and across multiple scenarios, characterized in that, The method includes: Acquire marine environmental data, ship attitude data, and historical trajectory data of preset targets; Wind and wave data are generated based on the marine environment data, and the wind and wave data is used to characterize the impact of wind and waves on the hull roll and the motion of the preset target. The historical trajectory data is processed to form the motion trend trajectory of the preset target; By combining the wind and wave data with the motion trend track, the simulated position of the preset target is corrected, and the pre-trajectory of the preset target is generated; The pre-trajectory is converted into a pre-pointing quantity corresponding to the acquisition device, and attitude compensation is performed in combination with the ship attitude data to form the turntable control quantity of the acquisition device. The acquisition device is controlled to acquire data from the preset target based on the turntable control quantity.

2. The method for collecting data from multiple scenarios in all weather conditions on a ship according to claim 1, characterized in that, The process of processing the historical trajectory data to form the motion trend trajectory of the preset target specifically includes: The historical trajectory data is rearranged according to the collection time, and the azimuth, elevation, distance, collection channel identifier and target confidence of the same preset target in the historical trajectory data are merged to form a target sequence. The target points in the target sequence are preprocessed to form a corrected target sequence; The bearing, pitch, and distance in the corrected target sequence are converted into target positions in the ship's coordinate system, and the continuous target positions are smoothed to form a smooth position sequence. Based on the smooth position sequence, the target velocity, target acceleration, and turning trend of the preset target are determined, and correlated according to the acquisition time to form the motion trend trajectory.

3. The method for collecting data from multiple scenarios in all weather conditions on a ship according to claim 1, characterized in that, The step of combining the wind and wave data with the motion trend trajectory to correct the simulated position of the preset target and generate the preset trajectory of the preset target specifically includes: Extract motion trend points from the motion trend trajectory; Based on the motion trend points, determine the smooth position, target velocity, target acceleration, turning trend, and acquisition channel identifier of the preset target; Based on the smooth position, the target velocity, the target acceleration, and the turning trend, the motion position of the preset target after the prediction time interval is extrapolated to form the simulated position; Wind disturbance vector, wave disturbance vector and wave period are extracted from the wind and wave data to generate wind and wave correction. The wind and wave correction amount is superimposed on the simulated position to form the corrected position of the preset target, and the corrected position is verified by combining the acquisition channel identifier; Multiple corrected positions are generated sequentially according to multiple prediction time intervals, and the multiple corrected positions are connected in chronological order to form the pre-trajectory.

4. The method for collecting data from multiple scenarios in all weather conditions on a ship according to claim 1, characterized in that, The data acquisition device includes a turntable mechanism, which specifically includes: A hull connecting seat is provided at the hull installation position, and the hull connecting seat forms the installation reference of the turntable mechanism relative to the hull coordinate system. An azimuth seat is provided on the hull connecting seat. The azimuth seat includes an azimuth axis system, an azimuth motor, an azimuth encoder, and a slip ring. The azimuth motor drives the azimuth axis system to rotate based on the turntable control quantity. The azimuth encoder provides feedback on the azimuth angle of the azimuth axis system. The slip ring is used to transmit signals when the azimuth axis system rotates continuously. A roll compensation ring is set between the azimuth mount and the pitch support. The roll compensation ring includes a roll motor, a roll encoder and a roll gyroscope. The roll motor drives the roll compensation ring to perform reverse compensation based on the roll angle in the hull attitude data. The roll encoder provides feedback on the roll compensation angle and the roll gyroscope provides feedback on the roll angular velocity. The pitch support is mounted on the roll compensation ring. The pitch support includes a pitch axis system, a pitch motor, a pitch encoder, and a pitch limiting component. The pitch motor drives the pitch axis system to rotate based on the turntable control quantity. The pitch encoder provides feedback on the pitch angle of the pitch axis system. The pitch limiting component is used to limit the rotation range of the pitch axis system. The carrier cabin is mounted on the pitch support, and a common reference load frame is provided inside the carrier cabin. The common reference load frame is used to fix the multi-mode camera and keep the multi-mode camera in a fixed line-of-sight correspondence. The fine-tuning frame is set in the carrier cabin. The fine-tuning frame is used to correct the angle of the line of sight of the multi-mode camera after the orientation seat, the roll compensation ring and the pitch support have completed the pointing compensation. A pre-aiming control chamber is installed inside the azimuth mount or on the fixed side of the azimuth mount. The pre-aiming control chamber converts the pre-aiming quantity into sub-control quantities that drive the azimuth mount, the roll compensation ring, the pitch support, and the fine-tuning frame, respectively.

5. The method for collecting data from multiple scenarios in all weather conditions on a ship according to claim 4, characterized in that, The process of converting the pre-trajectory into a pre-pointing quantity corresponding to the acquisition device, and combining it with the ship attitude data for attitude compensation to form the turntable control quantity of the acquisition device, specifically includes: Read the pre-trajectory points corresponding to the current control cycle from the pre-trajectory, and convert the pre-trajectory points into the relative position of the preset target relative to the acquisition device based on the installation reference; Based on the bow angle, pitch angle and roll angle in the hull attitude data, attitude compensation is performed on the relative position to form the compensated position of the preset target relative to the acquisition device; The pre-pointing value corresponding to the acquisition device is determined based on the compensation position. The pre-pointing value includes a pre-azimuth angle for driving the azimuth seat, a pre-pitch angle for driving the pitch support, and a pre-roll angle for driving the roll compensation ring. The pre-pointing value is input into the pre-aiming control chamber, and combined with the azimuth angle fed back by the azimuth encoder, the pitch angle fed back by the pitch encoder, the roll compensation angle fed back by the roll encoder, and the roll angular velocity fed back by the roll gyroscope, the turntable control value is generated.

6. The method for collecting data from multiple scenarios in all weather conditions on a ship according to claim 5, characterized in that, The method further includes: The difference between the pre-azimuth angle and the azimuth angle is processed to obtain the azimuth deviation; Based on the azimuth deviation and azimuth fine-tuning threshold, an azimuth sub-control quantity is generated, which drives the azimuth motor to rotate the azimuth axis system. The difference between the pre-roll angle and the roll compensation angle is processed to obtain the roll deviation, and the roll angular velocity is combined to generate the roll control quantity, so that the roll control quantity drives the roll motor to drive the roll compensation ring to perform reverse compensation. The pitch deviation is obtained by processing the difference between the pre-pitch angle and the pitch angle, and a pitch sub-control quantity is generated based on the pitch deviation and the pitch fine-tuning threshold. The pitch sub-control quantity drives the pitch motor to rotate the pitch axis. Based on the remaining portion of the azimuth deviation not compensated by the azimuth mount, the remaining portion of the pitch deviation not compensated by the pitch support, the remaining line-of-sight tilt after the reverse compensation of the roll compensation ring, and the line-of-sight calibration table corresponding to the multi-mode camera, a fine-tuning control quantity is generated, which drives the fine-tuning frame to correct the angle of the line-of-sight of the multi-mode camera. After the azimuth mount, the roll compensation ring, the pitch support, and the fine-tuning frame complete their actions, the residual deviation is determined based on the feedback results from the azimuth encoder, the roll encoder, the roll gyroscope, the pitch encoder, and the fine-tuning frame. Based on the residual deviation, the azimuth sub-control quantity, roll sub-control quantity, pitch sub-control quantity, and fine-tuning sub-control quantity corresponding to the next control cycle are updated.

7. The method for collecting data from multiple scenarios in all weather conditions on a ship according to claim 1, characterized in that, The step of controlling the acquisition device to acquire data from the preset target based on the turntable control quantity further includes: Based on the marine environmental data and the state of the preset target, a collection channel is determined from the multi-mode video of the collection device, wherein the multi-mode video includes visible light video, low-light video, or infrared video; The preset target is collected through the acquisition channel.

8. A shipboard all-weather, multi-scenario data collection device, characterized in that, The device is used to execute a ship all-weather multi-scenario data acquisition method as described in any one of claims 1-7, the device comprising an acquisition module, a processing module, and an output module, wherein: The acquisition module is used to acquire marine environmental data, ship attitude data, and historical trajectory data of preset targets; The processing module is used to generate wind and wave data based on the marine environment data, and the wind and wave data is used to characterize the impact of wind and waves on the hull roll and the motion of the preset target; The processing module is used to process the historical trajectory data to form the motion trend trajectory of the preset target; The processing module is used to combine the wind and wave data with the motion trend track to correct the simulated position of the preset target and generate the pre-trajectory of the preset target; The processing module is used to convert the pre-trajectory into a pre-pointing quantity corresponding to the acquisition device, and combine it with the ship attitude data to perform attitude compensation, thereby forming the turntable control quantity of the acquisition device. The output module is used to control the acquisition device to acquire data from the preset target based on the turntable control quantity.

9. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a computer program, wherein the computer program, when executed by the processor, implements a ship all-weather multi-scene data acquisition method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform a ship all-weather multi-scenario data acquisition method as described in any one of claims 1-7.

Images