A shipborne photoelectric detection integrated turntable device and a photoelectric detection method thereof

Abstract

The present application relates to the technical field of photoelectric detection, in particular to a shipborne photoelectric detection integrated turntable device and a photoelectric detection method thereof. The present application firstly acquires the image plane tailing amount in a single frame image exposure window, and maps the rotation posture at the exposure starting moment into a two-dimensional addressing coordinate; then, according to the deviation of the image plane tailing amount corresponding to the two-dimensional addressing coordinate relative to the image plane expected tailing amount corresponding to the same addressing coordinate and the adjacent addressing coordinate, the present application acquires the interference discrimination coefficient and updates the preset statistical table; finally, before the exposure of the next frame image is opened, the present application queries the preset statistical table by using the estimated rotation posture, and adaptively adjusts the exposure time length and the observation weight of the tracking filter. The present application discriminates the interference based on the index of the turntable rotation state in the preset statistical table, blocks the pollution of external impact on the mechanical degradation statistics; according to the index result of the predicted rotation posture, the present application adaptively adjusts the detection process, and improves the target detection effect in severe sea conditions.

Description

Technical Field

[0001] This invention relates to the field of photoelectric detection technology, specifically to a shipborne integrated photoelectric detection turntable device and its photoelectric detection method. Background Technology

[0002] Shipborne optoelectronic detection is a core piece of equipment for maritime target surveillance and precise tracking. To cope with the rolling of the ship caused by complex sea conditions, an integrated servo turntable is usually used to maintain stable alignment of the optical axis of view. In actual tracking operations, the servo motor inside the turntable needs to perform high-frequency commutation to counteract external disturbances in real time. When the motor crosses the zero speed point, the gear meshing gap causes mechanical vibration. If the vibration falls into the exposure window of the optoelectronic camera, it will cause the target spot to slide on the image plane, producing pixel trailing and blurring, causing the extracted target coordinates to jump.

[0003] Existing technologies typically employ fixed compensation mapping tables or moving average filtering to post-process and suppress the aforementioned errors. However, turntable gears undergo irreversible progressive wear during long-term service, and fixed mapping tables cannot track this gradual degradation pattern. Simultaneously, occasional extreme wind and wave impacts at sea can superimpose large transient disturbances. Conventional filtering algorithms cannot effectively distinguish between internal steady-state wear and sudden external impacts, leading to isolated extreme wave values ​​being incorrectly mixed into historical statistics, contaminating the compensation baseline, and consequently outputting distorted weights to the image tracking filter, ultimately causing target track deviations or even loss. Summary of the Invention

[0004] To address the technical problem of poor performance in existing shipborne photoelectric detection systems, the present invention aims to provide an integrated shipborne photoelectric detection turntable device and its photoelectric detection method. The specific technical solution adopted is as follows: A photoelectric detection method for a shipborne photoelectric detection integrated turntable device, the method comprising: The rotation attitude sequence of the turntable within the window during the exposure of a single frame image is extracted, and the image plane trailing amount of the target within the exposure window is calculated by combining optical parameters. Extract the rotation attitude at the start of the exposure window of the current frame image and quantize it into two-dimensional addressing coordinates of the turntable rotation attitude; in the preset statistics table, based on the image plane trailing amount within the exposure window corresponding to the two-dimensional addressing coordinates, and the deviation of the expected image plane trailing amount corresponding to the same addressing coordinates and adjacent addressing coordinates, obtain the interference discrimination coefficient, and update the preset statistics table based on the interference discrimination coefficient. Before the next frame of the image is exposed, the expected image trailing amount is obtained by querying a preset statistical table using the estimated rotation posture, and the exposure time and tracking filter are adaptively adjusted based on the expected image trailing amount.

[0005] Furthermore, the rotational attitude sequence is a time-series sequence of the turntable's angular velocity, and its acquisition method includes: The angular velocity time series of the turntable within the exposure window is obtained by interpolating the continuously acquired turntable angular velocities.

[0006] Furthermore, the method for obtaining the image plane trailing amount includes: The rotational attitude sequence is integrated temporally with the value direction removed to obtain the cumulative offset of the detection axis; the cumulative offset is converted into a physical displacement on the photosensitive plane based on the focal length, and the physical displacement is converted into an image plane trailing amount based on the pixel size.

[0007] Furthermore, the method for obtaining the two-dimensional addressing coordinates includes: Extract the angular velocity at the start of the exposure window of the current frame image, and calculate the angular acceleration at the start of the exposure window based on the angular velocity; perform a rounding operation on the angular velocity and the angular acceleration at the resolution of a preset statistical table, and perform out-of-bounds clamping on the rounding results to generate two-dimensional addressing coordinates in the preset statistical table.

[0008] Furthermore, the method for obtaining the interference discrimination coefficient includes: Extract the adjacent addressing coordinates of the two-dimensional addressing coordinates in a preset neighborhood of a preset statistical table, and determine the image plane reference trailing amount of the two-dimensional addressing coordinates based on the statistical characteristics of the image plane expected trailing amount corresponding to the adjacent addressing coordinates. The deviation of the image plane trailing amount from the expected image plane trailing amount is normalized to obtain a first discrimination parameter; the deviation of the image plane trailing amount from the reference image plane trailing amount is normalized to obtain a second discrimination parameter; the two discrimination parameters are fused, and the fusion result is numerically clamped, and the clamping result is used as the interference discrimination coefficient; If the two-dimensional addressing coordinates are not marked with the expected image trailing amount in the preset statistical table, then the two-dimensional addressing coordinates are marked as inactive. At the same time, if neither of its adjacent addressing coordinates is marked with the expected image trailing amount, then the interference discrimination coefficient is directly set to 0.

[0009] Furthermore, the preset neighborhood is an 8-neighborhood.

[0010] Furthermore, the method for updating the preset statistical table based on the interference discrimination coefficient includes: If the interference discrimination coefficient is 0 and the image plane trailing amount of the two-dimensional addressing coordinates does not exceed the preset tolerance limit, the image plane trailing amount is directly filled into the corresponding addressing coordinates in the preset statistics table as the expected image plane trailing amount, and the two-dimensional addressing coordinates are modified to the active state; when the preset tolerance limit is exceeded, the two-dimensional addressing coordinates remain in the inactive state. If the interference discrimination coefficient is not 0, based on the deviation of the image plane trailing amount of the two-dimensional addressing coordinates from the expected image plane trailing amount in the preset statistical table, and the interference discrimination coefficient, update the expected image plane trailing amount of the two-dimensional addressing coordinates in the preset statistical table.

[0011] Furthermore, the method for updating the expected image plane trailing amount of the two-dimensional addressing coordinates in a preset statistical table includes: The deviation of the image plane trailing amount of the two-dimensional addressing coordinates relative to the image plane expected trailing amount in the preset statistical table is calculated. The update weight is determined based on the negative correlation normalization value of the interference discrimination coefficient. The deviation is weighted using the update weight, and the weighted result is superimposed with the image plane expected trailing amount to obtain the updated image plane expected trailing amount.

[0012] Furthermore, methods for adaptively adjusting exposure time and tracking filters based on expected image trailing include: If the expected image trailing exceeds the preset tolerance limit, the exposure time will be adjusted to the preset lower limit; otherwise, the exposure time will be restored to the preset normal exposure time. The expected image plane trailing amount is weighted using a preset penalty weight to obtain an amplification weight. The amplification weight is then used to weight the main diagonal elements in the measurement noise covariance matrix of the Kalman filter to update the Kalman filter.

[0013] A shipborne integrated photoelectric detection turntable device, wherein the device implements the steps of the photoelectric detection method of the shipborne integrated photoelectric detection turntable device.

[0014] The present invention has the following beneficial effects: This invention extracts the rotational attitude sequence of the turntable within the exposure window of a single frame image and calculates the image plane trailing amount of the detected target within the exposure window by combining optical parameters, preparing for subsequent updates to the preset statistical table and adjustments to the detection. It extracts the rotational attitude at the start of the current frame image exposure window and quantizes and maps it to two-dimensional addressing coordinates of the turntable rotational attitude, providing a basis for subsequent addressing, extraction, or updating of the expected image plane trailing amount in the preset statistical table. Then, in the preset statistical table, based on the image plane trailing amount within the exposure window corresponding to the two-dimensional addressing coordinates, and comparing it with the expected image plane trailing amount corresponding to the same and adjacent addressing coordinates, it analyzes the degree of trailing deviation under similar historical rotational attitudes, thereby removing external interference, obtaining interference discrimination coefficients, and updating the preset statistical table based on the interference discrimination coefficients, while retaining normal electromechanical degradation data of the turntable to ensure the reference reliability of the statistical table. Before the next frame image exposure begins, it uses the estimated rotational attitude to query the preset statistical table to obtain the expected image plane trailing amount, and adaptively adjusts the exposure time and tracking filter based on the expected image plane trailing amount. This invention uses the rotation state of the turntable to index the data in a preset statistical table for interference discrimination, thus blocking the contamination of internal mechanical degradation statistics by extreme wave impacts. Furthermore, it adaptively adjusts the exposure time and filter observation weights during the detection process based on the index results of the preset statistical table, thereby improving the continuity and stability of target detection under severe sea conditions. Attached Figure Description

[0015] Figure 1 This is a flowchart of a photoelectric detection method for a shipborne photoelectric detection integrated turntable device provided in one embodiment of the present invention. Detailed Implementation

[0016] The following description, in conjunction with the accompanying drawings, details the specific scheme of the shipborne photoelectric detection integrated turntable device and its photoelectric detection method provided by the present invention.

[0017] Please see Figure 1 The diagram illustrates a flowchart of a shipborne integrated photoelectric detection turntable device and its photoelectric detection method according to an embodiment of the present invention, specifically including: Step S1: Extract the rotation attitude sequence of the turntable within the window during the exposure of a single frame image, and calculate the image plane trailing amount of the target within the exposure window by combining optical parameters.

[0018] The mechanical state changes caused by gear backlash degradation and wave impact on the shipborne integrated photoelectric detection turntable are the physical root cause of the deviation of the photoelectric detection line of sight. Therefore, the embodiment of the present invention first collects the rotation attitude of the turntable to prepare for subsequent quantification of the actual mechanical jitter amplitude that leads to the decrease in photoelectric detection accuracy. Furthermore, since the light spot trailing phenomenon of the detected target during the photoelectric detection process occurs during the absolute physical period when the detector shutter is open, the mechanical vibration during the non-exposure period will not affect the actual detection results. Therefore, by extracting the rotation attitude sequence of the turntable within the exposure window of a single frame image, it can be ensured that there is a precise causal mapping between the mechanical jitter amplitude and the error performance of the current photoelectric detection, providing a basis for accurately measuring the amount of image plane trailing of the detected target within the exposure window.

[0019] Taking any single-frame image as an example, first determine the exposure window of the single-frame image; continuously monitor the shutter enable pin level transition at the bottom of the photodetector, extract the actual shutter opening timestamp ts and the actual shutter closing timestamp te for each frame, and construct the exposure window [ts,te] of the single-frame image.

[0020] In a preferred embodiment of the present invention, considering that the electromechanical vibration of the turntable is essentially an angular motion of the line of sight occurring in a very short time, and that the angular velocity output by the underlying fiber optic gyroscope is the most direct and sensitive physical quantity characterizing this motion, the rotation attitude sequence is defined as the angular velocity time sequence of the turntable. Furthermore, considering that the angular velocity is a discrete sampling point continuously output by the fiber optic gyroscope at a fixed frequency, while the opening and closing of the shutter of the photoelectric camera is an absolute physical action triggered by the outside, the occurrence times of the two are highly unlikely to be strictly aligned, and directly extracting the internal sampling points will lead to time domain truncation errors at both ends; therefore, the angular velocity time sequence of the turntable within the exposure window is obtained by interpolation based on the continuously acquired turntable angular velocity.

[0021] As an example, access the underlying data buffer of the inner-loop fiber optic gyroscope of the turntable, read the angular velocity sampling points output by the gyroscope with a set sampling frequency such as 1000Hz; for the start timestamp ts, extract the two angular velocity sampling points immediately before and after ts on the time axis, and then calculate the (interpolated) angular velocity at time ts using a linear interpolation algorithm; similarly, calculate the (interpolated) angular velocity at time te; extract all angular velocity sampling points in the interval [ts, te], and sort them in chronological order to form an angular velocity time sequence aligned with the exposure window.

[0022] The light spot trailing caused by the electromechanical vibration of the turntable is physically due to the continuous spatial sliding of the light spot relative to the photosensitive chip during exposure. The spatial range of this sliding directly determines the geometric degree of image blur. However, the electromechanical vibration described based on the turntable rotation angle cannot intuitively reflect the image plane trailing situation. Therefore, this embodiment of the invention further calculates the image plane trailing amount of the detection target within the exposure window based on the rotation attitude sequence of the turntable and combined with optical parameters; the image plane trailing amount quantifies the mechanical angle jitter of the bottom layer of the turntable into an image blur scale at the photoelectric imaging level, which prepares for subsequent updates to the preset statistical table and adjustments to the detection.

[0023] Preferably, in one embodiment of the present invention, considering that the turntable may generate high-frequency alternating vibrations in both directions, if the direction is directly integrated with respect to the angular velocity, the angular displacements in both directions will algebraically cancel each other out, masking the real trailing effect caused by the repeated friction of the light spot on the photosensitive plane; while stripping the absolute value integration of the rotation attitude sequence to force all the reverse vibrations to be converted into a positive cumulative path, it is possible to accurately restore the absolute cumulative offset generated by the detection axis during exposure; considering that the focal length of the optical system directly determines the amplification effect of the angular deviation on the imaging plane, the cumulative offset can be converted into a physical displacement on the photosensitive plane based on the focal length, and then the physical displacement can be quantified into image plane trailing amount by combining the pixel size, providing a digital measurement basis for subsequently measuring the degree of degradation of photoelectric images; the method for obtaining image plane trailing amount includes: The rotational attitude sequence is integrated temporally to remove the value direction, and the cumulative offset of the detection axis is obtained. Based on the focal length, the cumulative offset is converted into a physical displacement on the photosensitive plane, and based on the pixel size, the physical displacement is converted into image plane trailing.

[0024] As an example, extract all angular velocities in the rotational attitude sequence and the time difference between adjacent angular velocity sampling points; in chronological order, take the absolute value of the angular velocity at two adjacent moments, calculate the mean of the two absolute values, and multiply it by the corresponding time difference to obtain the absolute angular displacement of the probe line of sight within a single time slice corresponding to adjacent angular velocity sampling points; sum up the absolute angular displacements of the probe line of sight within a single time slice corresponding to all adjacent angular velocity sampling points to obtain the cumulative offset of the probe line of sight within the entire exposure window; The focal length of the optical lens configured by the photoelectric detection equipment is retrieved, and the cumulative offset of the detection axis is multiplied by the focal length of the optical lens to calculate the actual physical displacement of the target spot on the photosensitive plane. Then, the physical displacement in angular form is converted into physical displacement in radian form. The physical size of a single pixel on the photosensitive plane is retrieved, and the physical displacement in radian form is divided by the size of the single pixel to calculate the image plane trailing amount (i.e., the actual number of pixels that the spot crosses in the image).

[0025] Step S2: Extract the rotation attitude at the start of the exposure window of the current frame image and quantize it into two-dimensional addressing coordinates of the turntable rotation attitude; in the preset statistics table, according to the image plane trailing amount within the exposure window corresponding to the two-dimensional addressing coordinates, and the deviation of the expected image plane trailing amount corresponding to the same addressing coordinates and adjacent addressing coordinates, obtain the interference discrimination coefficient, and update the preset statistics table based on the interference discrimination coefficient.

[0026] It should be noted that, due to the complex electromechanical conditions faced by the shipborne photoelectric detection integrated turntable, the embodiments of the present invention pre-construct a structured storage container to classify and record the detection deviation (image plane trailing characteristics) under different operating conditions; specifically, a preset statistical table is constructed, which is composed of a discrete two-dimensional grid matrix. Each independent grid (i.e., storage unit) is used to archive and record the expected image plane trailing amount of the turntable under specific electromechanical conditions, and the grid index address is the two-dimensional addressing coordinate.

[0027] Because shipborne turntables exhibit significant differences in gear meshing states and mechanical vibration patterns under various electromechanical stress states (such as low-speed reversing, high-speed cruising, rapid acceleration, or rapid deceleration), a single global compensation parameter cannot accurately characterize the mechanical degradation under these variable operating conditions. Therefore, this embodiment of the invention extracts the specific rotational posture at the start of exposure for each frame of image, uses the instantaneous posture to finely characterize the electromechanical conditions of the turntable, and further quantizes and maps it into two-dimensional addressing coordinates of the turntable's rotational posture, providing a basis for subsequent addressing, extraction, or updating of the desired trailing amount of the image plane in a preset statistical table.

[0028] It should be noted that the shipborne integrated photoelectric detection turntable will generate a continuous sequence of photoelectric image frames during continuous operation, and the analysis and processing logic for all image frames is consistent; here, only the current frame is used as an example for analysis and description, and will not be elaborated on in detail.

[0029] Preferably, in one embodiment of the present invention, since the electromechanical force posture of the turntable is jointly determined by the speed of motion (angular velocity) and the urgency of the force (angular acceleration), the angular velocity at the start of exposure is extracted simultaneously and its angular acceleration is calculated to construct two-dimensional state information reflecting the transient force characteristics of the turntable. Furthermore, since both angular velocity and angular acceleration are continuously floating high-precision physical quantities, directly using them as memory indexes would lead to infinite expansion of storage space, and small fluctuations would not have classification significance. Simultaneously, angular velocity and angular acceleration may exceed the factory-set normal operating limits of the turntable, potentially leading to out-of-bounds errors and system crashes. Therefore, firstly, dimensionality reduction and rounding operations are performed on these two parameters using the step size resolution of a preset statistical table, forcibly converging and quantizing the continuous physical analog quantities into discrete integer indices. Then, out-of-bounds clamping is forcibly executed to ensure the absolute safety of the addressing coordinates of the statistical archive, thereby matching the addressing rules in the preset statistical table. The method for obtaining two-dimensional addressing coordinates includes: Extract the angular velocity at the start of the exposure window of the current frame image, and calculate the angular acceleration at the start of the exposure window based on the angular velocity; perform rounding operation on the angular velocity and angular acceleration at the resolution of the preset statistical table, and perform out-of-bounds clamping on the rounding results to generate two-dimensional addressing coordinates in the preset statistical table.

[0030] As an example, the angular velocity at the start of the exposure window and the immediately preceding angular velocity on the time axis are called; the difference between the starting angular velocity and the preceding angular velocity is divided by the absolute time difference between the two acquisition times corresponding to the two angular velocities to obtain the angular acceleration at the start. Retrieve the factory-set physical limit operating parameters of the turntable and servo motor. For example, the angular velocity limit is the maximum rated rotational speed of the turntable (e.g., 60 degrees / second), and the angular acceleration limit is the maximum rated angular acceleration of the servo motor (e.g., 120 degrees / square second). Then, based on the standard deviation of the noise floor of the fiber optic gyroscope under the turntable's powered-on and stationary conditions, set the step resolution of the angular velocity and the step resolution of the angular acceleration. For example, if the standard deviation of the noise floor σ of the fiber optic gyroscope is measured, then the step resolution of the angular velocity is set to 3σ, and the step resolution of the angular acceleration is... The quotient obtained by dividing σ by the sampling time interval; Among them, taking 3σ is based on the statistical 3σ principle to prevent the addressing coordinates from frequently changing due to random noise from the sensor. Based on the law of error propagation; angular acceleration is obtained by subtracting two adjacent angular velocities. Subtracting two independent, identically distributed random variables (both with standard deviation σ), the standard deviation of the difference becomes... σ, therefore take σ is divided by the sampling time interval to unify the dimensions.

[0031] Add the angular velocity limit to the initial angular velocity to eliminate the negative value effect of translation, then divide the sum by the step resolution of the angular velocity, and call the floor operator to truncate the decimal part to generate the initial horizontal coordinate; similarly, divide twice the angular velocity limit by the step resolution of the angular velocity, and call the floor operator to truncate the decimal part to generate the maximum horizontal coordinate index. The angular acceleration at the initial moment is added to the angular acceleration limit to eliminate the negative value effect through translation. Then, the sum is divided by the step size resolution of the angular acceleration, and the floor function is called to truncate the decimal part to generate the initial ordinate. Similarly, twice the angular acceleration limit is divided by the step size resolution of the angular acceleration, and the floor function is called to truncate the decimal part to generate the maximum ordinate index. If the initial x-coordinate is less than 0, it is forcibly assigned a value of 0; if it is greater than the maximum x-coordinate index, it is forcibly assigned the maximum x-coordinate index to perform out-of-bounds clamping and determine the final x-coordinate of the two-dimensional addressing coordinates. Similarly, if the initial y-coordinate is less than 0, it is forcibly assigned a value of 0; if it is greater than the maximum y-coordinate index, it is forcibly assigned the maximum y-coordinate index to perform out-of-bounds clamping and determine the final y-coordinate of the two-dimensional addressing coordinates. The final x-coordinate and the final y-coordinate together determine the two-dimensional addressing coordinates. Since the errors of the shipboard turntable are intertwined with the gradual degradation of internal gears and the impact of sudden external waves, and conventional filtering is very easy to absorb the extreme values ​​of wave impact, which leads to the distortion of historical compensation data, it is also necessary to obtain the interference discrimination coefficient to quantitatively evaluate and block external extreme interference, and only retain the real mechanical degradation data. Since the trailing amount corresponding to the same addressing coordinates reflects the long-term wear pattern in the time dimension, and the trailing amount corresponding to adjacent addressing coordinates reflects the smooth transition pattern in the spatial dimension, cross-validation by combining the same addressing coordinates and adjacent addressing coordinates can help to accurately isolate isolated external wave extremes, retain normal electromechanical degradation data of the turntable, and ensure the purity of the statistical table data. Based on this, in the embodiments of the present invention, the interference discrimination coefficient is obtained in the preset statistical table according to the image plane trailing amount within the exposure window corresponding to the two-dimensional addressing coordinates, relative to the expected image plane trailing amount corresponding to the same addressing coordinates and adjacent addressing coordinates. The interference discrimination coefficient characterizes the severity and probability of the image plane trailing error of the current frame originating from external sudden environmental interference (such as wave impact), providing a reliable criterion for subsequent updating of the preset statistical table and control detection.

[0032] Preferably, in one embodiment of the present invention, considering that the mechanical wear inside the turntable servo motor has physical continuity under similar stress postures, and that isolated wave impacts usually do not have spatial smoothness, the same addressing coordinates in the preset statistical table and adjacent addressing coordinates in the preset neighborhood are extracted from the two-dimensional addressing coordinates, and a reliable spatial statistical benchmark is provided based on the corresponding expected image plane trailing amount, thereby cross-validating the discrimination information from two perspectives, and clamping the fusion result to quantify the interference discrimination coefficient; considering that when the turntable servo motor is cold-started or experiences a specific stress posture for the first time, the corresponding addressing coordinates in the preset statistical table may be in an inactive state, that is, the image plane trailing amount is not recorded (historical), and the expected reference cannot be provided, so it is necessary to first check, and if it is in an inactive state, the interference discrimination coefficient is directly set to 0; the method for obtaining the interference discrimination coefficient includes: Extract the adjacent addressing coordinates of the two-dimensional addressing coordinates in the preset neighborhood of the preset statistical table, and determine the image plane reference trailing amount of the two-dimensional addressing coordinates based on the statistical characteristics of the image plane expected trailing amount corresponding to the adjacent addressing coordinates. The deviation of the image plane trailing amount from the expected image plane trailing amount is normalized to obtain the first discrimination parameter; the deviation of the image plane trailing amount from the reference image plane trailing amount is normalized to obtain the second discrimination parameter; the two discrimination parameters are fused, and the fusion result is numerically clamped, and the clamping result is used as the interference discrimination coefficient. If the two-dimensional addressing coordinates are not marked with the expected image trailing amount in the preset statistics table, then the two-dimensional addressing coordinates are marked as inactive. At the same time, if neither of its adjacent addressing coordinates is marked with the expected image trailing amount, the interference discrimination coefficient is directly set to 0.

[0033] As an example, all adjacent addressing coordinates of two-dimensional addressing coordinates within a preset neighborhood in a preset statistical table are extracted. In a preferred embodiment of the present invention, the preset neighborhood is an 8-neighborhood. If the same addressing coordinates and adjacent addressing coordinates in the preset statistics table do not have the expected image plane trailing amount marked, that is, they are all in the initial state, then the interference discrimination coefficient is directly set to 0; otherwise, the expected image plane trailing amount corresponding to the adjacent addressing coordinates is extracted from the preset statistics table (excluding the adjacent addressing coordinates in the initial state), and the mean is calculated as the image plane reference trailing amount of the two-dimensional addressing coordinates. The deviation is measured by the absolute value of the difference. The absolute value of the difference between the image plane trailing amount (calculated) under two-dimensional addressing coordinates and the image plane expected trailing amount corresponding to the same addressing coordinates in the preset statistical table is calculated. The absolute value of the difference is normalized to obtain the first discrimination parameter. The absolute value of the difference between the image plane trailing amount calculated under two-dimensional addressing coordinates and the image plane reference trailing amount is calculated. The absolute value of the difference is normalized to obtain the second discrimination parameter. The normalization method described above involves constructing a normalization denominator and normalizing by dividing by the normalization denominator. Specifically, a maximum trailing limit is set, for example, to be 30, representing the maximum trailing amount under interference conditions. A preset multiplier, such as 2, is configured and multiplied by the expected trailing amount of the image plane corresponding to the two-dimensional addressing coordinates in the preset statistical table. The maximum value of the maximum trailing limit and the product is taken as the normalization denominator. Implementers can adjust the normalization denominator according to actual applications or use other normalization methods. The first and second discriminant parameters are further weighted with weights of 0.5 and 0.5 respectively, and the weighted sum is clamped to within 1 to obtain the interference discrimination coefficient. The implementer may also use other fusion methods, such as addition or multiplication.

[0034] The interference discrimination coefficient accurately measures the probability of external extreme environmental impacts. Therefore, this coefficient is used to generate the update weights of the preset statistical table in reverse. This allows the update weights to automatically decay when encountering isolated wave impacts (interference discrimination coefficient approaches 1), thereby blocking the writing to the statistical table. However, during normal gradual degradation of machinery (interference discrimination coefficient approaches zero), the update weights remain high, thus allowing updates. Updating the preset statistical table based on this mechanism can eliminate external environmental pollution while ensuring the dynamic updating and writing of electromechanical wear characteristics. Based on this, the embodiments of the present invention will update the preset statistical table based on the interference discrimination coefficient, providing reliable data support for subsequent predictions.

[0035] Preferably, in one embodiment of the present invention, considering that during the first operation under a specific stress posture (the interference discrimination coefficient is forcibly set to 0), there is no historical data as a smoothing benchmark. Simultaneously, if the image plane trailing amount does not exceed the preset tolerance limit, it indicates that the initially measured image plane trailing amount can accurately reflect the current initial mechanical clearance of the equipment. It can be directly filled with the desired image plane trailing amount and the coordinates activated, thereby completing the initial baseline construction. Otherwise, it is discarded to avoid contaminating the statistical table. Furthermore, considering that in the two-dimensional addressing coordinate activation state, simply overwriting new values ​​would erase the long-term accumulated steady-state wear characteristics of the equipment, while the interference discrimination coefficient accurately measures the probability that the current error belongs to environmental shock, a residual iterative update mechanism combining historical and current data can dynamically adjust the combination ratio, preserving mechanical smoothing degradation traces while filtering out external disturbances from interfering with the update process; therefore, the method for updating the preset statistical table based on the interference discrimination coefficient includes: If the interference discrimination coefficient is 0 and the image plane trailing amount of the two-dimensional addressing coordinates does not exceed the preset tolerance limit, the image plane trailing amount is directly filled into the corresponding addressing coordinates in the preset statistics table as the expected image plane trailing amount, and the two-dimensional addressing coordinates are modified to the active state; when the preset tolerance limit is exceeded, the two-dimensional addressing coordinates remain in the inactive state. If the interference discrimination coefficient is not 0, based on the deviation of the image plane trailing amount of the two-dimensional addressing coordinates from the expected image plane trailing amount in the preset statistical table, and the interference discrimination coefficient, update the expected image plane trailing amount of the two-dimensional addressing coordinates in the preset statistical table.

[0036] As an example, if the interference discrimination coefficient is 0, it is further determined whether the image plane trailing amount of the two-dimensional addressing coordinates exceeds the preset tolerance limit, that is, the maximum trailing limit, such as 30. If it does not exceed the limit, the image plane trailing amount is directly filled into the corresponding addressing coordinates in the preset statistics table as the expected image plane trailing amount, and the two-dimensional addressing coordinates are modified to the active state; if it exceeds the limit, it is not overwritten and remains inactive. If the interference discrimination coefficient is not zero, considering that the magnitude of the interference discrimination coefficient is proportional to the severity of the external wave impact, the negative correlation normalization value inversion logic of the interference discrimination coefficient is used to determine the update weight. This ensures that the absorption weight of the current image plane trailing amount is forcibly reduced under severe sea conditions, while the reference level is increased under normal wear conditions. This is then superimposed with the original image plane expected trailing amount to smoothly track and adaptively update the image plane expected trailing amount that characterizes the actual progressive wear characteristics of the gears inside the turntable. Based on this, in a preferred embodiment of the present invention, the method for updating the image plane expected trailing amount of the two-dimensional addressing coordinates in the preset statistical table includes: The deviation of the image plane trailing amount of the two-dimensional addressing coordinates from the image plane expected trailing amount in the preset statistical table is calculated. The update weight is determined based on the negative correlation normalization value of the interference discrimination coefficient. The deviation is weighted using the update weight, and the weighted result is superimposed with the image plane expected trailing amount to obtain the updated image plane expected trailing amount.

[0037] Specifically, the difference between the image plane trailing amount (calculated) in two-dimensional addressing coordinates and the expected image plane trailing amount in the preset statistical table is calculated (the sign is retained, and the sign represents the update direction); the interference discrimination coefficient is subtracted from the constant 1 to perform negative correlation normalization, and the negative correlation normalization value is multiplied by the preset forgetting factor to obtain the update weight; then the update weight is multiplied by the difference and weighted, and the weighted result is added to the expected image plane trailing amount in the preset statistical table to obtain the updated expected image plane trailing amount.

[0038] The preset forgetting factor determines the learning rate, balancing the smoothness of the statistical table features with the sensitivity to tracking current wear. The preset forgetting factor must be greater than 0 and much less than 1, with a recommended value of 0.05. Implementers can also adjust it themselves.

[0039] Step S3: Before the exposure of the next frame image is turned on, the expected image plane trailing amount is obtained by querying the preset statistics table using the estimated rotation posture, and the exposure time and tracking filter are adaptively adjusted based on the expected image plane trailing amount.

[0040] Because the shipborne optoelectronic turntable has a large inertia and mechanical continuity in its force-driven motion, the abrupt change in rotational attitude is extremely small in a very short time, and the speed and acceleration will not change abruptly. Therefore, the micro-jitter behavior that will occur during the exposure period can be inferred before the exposure of the next frame image begins, that is, the rotational attitude can be estimated. Then, by looking up the table, the expected image plane trailing caused by physical wear under the estimated rotational attitude can be determined, providing a basis for subsequent compensation.

[0041] As an example, in the extremely short instant before the hardware trigger signal for the next frame of image is sent to the photodetector, the instantaneous angular velocity output by the inner-loop fiber optic gyroscope is read through the low-level driver, and the instantaneous angular acceleration is calculated by backtracking the angular velocity at the previous sampling moment. The instantaneous angular velocity and instantaneous angular acceleration are used as an effective estimate of the rotation state during the exposure of the next frame. Obtain the target two-dimensional addressing coordinates under the rotational attitude corresponding to the instantaneous angular velocity and instantaneous angular acceleration. This is the same as the step of obtaining the two-dimensional addressing coordinates in step S2, and will not be repeated here. After determining the target two-dimensional addressing coordinates, query the preset statistical table to obtain the expected image plane trailing amount at the target two-dimensional addressing coordinates.

[0042] It should be noted that the expected image plane trailing amount may not have been recorded at the target two-dimensional addressing coordinates, i.e., it is in an inactive state. In this case, the detection process corresponding to the next frame image will not be adjusted, and the normal exposure time of the factory configuration, such as 20ms, will continue to be maintained. At the same time, the original basic measurement noise covariance matrix will remain unchanged.

[0043] It should be noted that, in this embodiment of the invention, the shipborne optoelectronic detection integrated turntable is equipped with a Kalman filter (tracking filter) for target trajectory tracking.

[0044] Because the expected image trailing amount accurately predicts the degree of blurring of the image in the window due to internal mechanical jitter during future exposure, when the turntable is about to cross the violent vibration range, the physical sliding length of the light spot on the photosensitive layer can be actively reduced by reducing the exposure time of the hardware camera, thereby sacrificing part of the total light intake to exchange for the sharpness of the target spatial coordinates. Furthermore, while adjusting the exposure time during the detection process suppresses motion blur, the decrease in imaging signal-to-noise ratio leads to random jumps in the target extraction coordinates. These jumps manifest as increased measurement noise under Kalman filtering. Traditional Kalman filters set the measurement noise covariance matrix to a constant value, failing to detect the degradation in observation quality caused by jitter. If the filter maintains fixed observation weights, the target track will deviate from its true state. However, before the Kalman filter performs state updates, it adaptively adjusts the observation weights of the tracking filter based on the expected image trailing amount. Essentially, this adjusts the confidence weight of the current detection information in state estimation, thereby suppressing the impact of turntable vibrations on target detection under complex sea conditions.

[0045] Based on this, embodiments of the present invention further adaptively adjust the exposure time and the tracking filter based on the expected image trailing amount.

[0046] Preferably, in one embodiment of the present invention, considering that when the expected image plane trailing amount exceeds the preset tolerance limit (the maximum blur threshold that image feature recognition can tolerate), adjusting the exposure time to a preset lower limit (such as the minimum allowable light sensitivity time of the camera) can cut off the physical sliding path of the light spot on the photosensitive chip to the greatest extent while ensuring basic imaging brightness; and considering that using preset penalty weights to weight the expected image plane trailing amount can help assess the unreliability of the frame image, and in the measurement noise covariance matrix (observation matrix) of the Kalman filter, the main diagonal elements represent the random noise variance level of each independent observation dimension (such as azimuth and elevation coordinates), while the off-diagonal elements represent the coupling characteristics between dimensions, weighting and amplifying only the main diagonal elements can improve the uncertainty assessment of the current observation result without destroying the factory coupling calibration, thereby automatically reducing the Kalman gain in the filtering calculation, forcing the algorithm to reduce the absorption weight of the current frame observation coordinates, so this mechanism adaptively dilutes the residual error on the software side and ensures the stability of the tracking track; The methods for adaptively adjusting the exposure time and tracking filter based on the expected image trailing amount include: If the expected image trailing exceeds the preset tolerance limit, the exposure time will be adjusted to the preset lower limit; otherwise, the exposure time will be restored to the preset normal exposure time. The expected image plane trailing amount is weighted using a preset penalty weight to obtain the amplification weight. The amplification weight is then used to weight the main diagonal elements in the measurement noise covariance matrix of the Kalman filter to update the Kalman filter.

[0047] As an example, the preset tolerance limit is set to 30, the preset lower limit of exposure time is 2ms, and the preset normal exposure time is 20ms. Implementers can also adjust these according to actual conditions. The preset penalty weight is set to 0.5. The expected image trailing amount is multiplied by the preset penalty weight, and then a constant of 1 is added to obtain the amplification weight. Then, the diagonal elements of the measurement noise covariance matrix of the Kalman filter are multiplied by the amplification weight and overwritten to update the measurement noise covariance matrix. Then, it is substituted into the standard state update process of the Kalman filter to update the detection state.

[0048] The present invention also proposes a shipborne integrated photoelectric detection turntable device, which implements the photoelectric detection method of the shipborne integrated photoelectric detection turntable device described in steps S1-S3.

[0049] In summary, this invention extracts the rotational attitude sequence of the turntable within the exposure window of a single frame image and calculates the image plane trailing amount of the detected target within the exposure window by combining optical parameters; it extracts the rotational attitude at the start of the exposure window of the current frame image and quantizes it into two-dimensional addressing coordinates of the turntable rotational attitude; in a preset statistical table, based on the image plane trailing amount corresponding to the two-dimensional addressing coordinates within the exposure window, and the deviation of the expected image plane trailing amount corresponding to the same addressing coordinates and adjacent addressing coordinates, it obtains the interference discrimination coefficient and updates the preset statistical table based on the interference discrimination coefficient; before the exposure of the next frame image begins, it uses the estimated rotational attitude to query the preset statistical table to obtain the expected image plane trailing amount, and adaptively adjusts the exposure time and tracking filter based on the expected image plane trailing amount. This invention uses the turntable rotation state to index the preset statistical table for interference discrimination, blocking the contamination of internal mechanical degradation statistics by extreme wave impacts; further, it adaptively adjusts the exposure time and filter observation weights during the detection process based on the index results of the preset statistical table, improving the continuity and stability of target detection under severe sea conditions.

Claims

1. A photoelectric detection method for a shipborne photoelectric detection integrated turntable device, characterized in that, The method includes: The rotation attitude sequence of the turntable within the window during the exposure of a single frame image is extracted, and the image plane trailing amount of the target within the exposure window is calculated by combining optical parameters. Extract the rotation attitude at the start of the exposure window of the current frame image and quantize it into two-dimensional addressing coordinates of the turntable rotation attitude; in the preset statistics table, based on the image plane trailing amount within the exposure window corresponding to the two-dimensional addressing coordinates, and the deviation of the expected image plane trailing amount corresponding to the same addressing coordinates and adjacent addressing coordinates, obtain the interference discrimination coefficient, and update the preset statistics table based on the interference discrimination coefficient. Before the exposure of the next frame image begins, the expected image trailing amount is obtained by querying a preset statistical table using the estimated rotation attitude, and the exposure time and tracking filter are adaptively adjusted based on the expected image trailing amount; the rotation attitude sequence is the angular velocity time sequence of the turntable, and its acquisition method includes: The turntable angular velocity time series within the exposure window is obtained by interpolating the continuously acquired turntable angular velocities. The method for obtaining the interference discrimination coefficient includes: Extract the adjacent addressing coordinates of the two-dimensional addressing coordinates in a preset neighborhood of a preset statistical table, and determine the image plane reference trailing amount of the two-dimensional addressing coordinates based on the statistical characteristics of the image plane expected trailing amount corresponding to the adjacent addressing coordinates. The deviation of the image plane trailing amount from the expected image plane trailing amount is normalized to obtain a first discrimination parameter; the deviation of the image plane trailing amount from the reference image plane trailing amount is normalized to obtain a second discrimination parameter; the two discrimination parameters are fused, and the fusion result is numerically clamped, and the clamping result is used as the interference discrimination coefficient; If the two-dimensional addressing coordinates are not marked with the expected image plane trailing amount in the preset statistics table, then the two-dimensional addressing coordinates are marked as inactive. At the same time, if neither of its adjacent addressing coordinates is marked with the expected image plane trailing amount, then the interference discrimination coefficient is directly set to 0. Methods for adaptively adjusting exposure time and tracking filters based on expected image trailing include: If the expected image trailing exceeds the preset tolerance limit, the exposure time will be adjusted to the preset lower limit; otherwise, the exposure time will be restored to the preset normal exposure time. The expected image plane trailing amount is weighted using a preset penalty weight to obtain an amplification weight. The amplification weight is then used to weight the main diagonal elements in the measurement noise covariance matrix of the Kalman filter to update the Kalman filter.

2. The photoelectric detection method of the shipborne photoelectric detection integrated turntable device according to claim 1, characterized in that, The method for obtaining the image plane trailing amount includes: The rotational attitude sequence is integrated temporally with the value direction removed to obtain the cumulative offset of the detection axis; the cumulative offset is converted into a physical displacement on the photosensitive plane based on the focal length, and the physical displacement is converted into an image plane trailing amount based on the pixel size.

3. The photoelectric detection method of the shipborne photoelectric detection integrated turntable device according to claim 1, characterized in that, The method for obtaining the two-dimensional addressing coordinates includes: Extract the angular velocity at the start of the exposure window of the current frame image, and calculate the angular acceleration at the start of the exposure window based on the angular velocity; perform a rounding operation on the angular velocity and the angular acceleration at the resolution of a preset statistical table, and perform out-of-bounds clamping on the rounding results to generate two-dimensional addressing coordinates in the preset statistical table.

4. The photoelectric detection method of the shipborne photoelectric detection integrated turntable device according to claim 1, characterized in that, The preset neighborhood is an 8-neighborhood.

5. The photoelectric detection method of the shipborne photoelectric detection integrated turntable device according to claim 1, characterized in that, Methods for updating preset statistical tables based on interference discrimination coefficients include: If the interference discrimination coefficient is 0 and the image plane trailing amount of the two-dimensional addressing coordinates does not exceed the preset tolerance limit, the image plane trailing amount is directly filled into the corresponding addressing coordinates in the preset statistics table as the expected image plane trailing amount, and the two-dimensional addressing coordinates are modified to the active state; when the preset tolerance limit is exceeded, the two-dimensional addressing coordinates remain in the inactive state. If the interference discrimination coefficient is not 0, based on the deviation of the image plane trailing amount of the two-dimensional addressing coordinates from the expected image plane trailing amount in the preset statistical table, and the interference discrimination coefficient, update the expected image plane trailing amount of the two-dimensional addressing coordinates in the preset statistical table.

6. The photoelectric detection method of the shipborne photoelectric detection integrated turntable device according to claim 5, characterized in that, The method for updating the expected image plane trailing amount of the two-dimensional addressing coordinates in a preset statistical table includes: The deviation of the image plane trailing amount of the two-dimensional addressing coordinates relative to the image plane expected trailing amount in the preset statistical table is calculated. The update weight is determined based on the negative correlation normalization value of the interference discrimination coefficient. The deviation is weighted using the update weight, and the weighted result is superimposed with the image plane expected trailing amount to obtain the updated image plane expected trailing amount.

7. A shipborne integrated photoelectric detection turntable device, characterized in that, The device implements the steps of the photoelectric detection method of the shipborne photoelectric detection integrated turntable device as described in any one of claims 1 to 6.

Images