A wide temperature self-adaptive continuous zoom television camera control method and system

Abstract

The present application relates to the field of automatic control technology, in particular to a kind of wide temperature self-adapting continuous zoom television camera control method and system, the method includes: obtaining the zoom focus mapping relationship of continuous zoom television camera, system inherent comprehensive time delay, mechanical inertia time delay, focal depth and the absolute position data of zoom motor, calculate transient motion parameter, determine dynamic lead displacement amount in combination with inherent comprehensive time delay;In mapping relationship, deduce reference and predicted focus position, determine defocus vulnerability in combination with focal depth;Extract zoom focus sensitivity, determine focus displacement compensation in combination with transient motion parameter and mechanical inertia time delay;Focus displacement compensation is weightedly corrected using defocus vulnerability and superimposed to predicted focus position, generate adaptive feedforward focus target coordinates.The method improves the focus point following precision in wide temperature working condition and high dynamic continuous zoom process.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology, specifically to a control method and system for a wide-temperature adaptive continuous zoom television camera. Background Technology

[0002] In fields such as border and coastal defense monitoring and aerospace reconnaissance, continuous zoom television cameras are often used as optoelectronic payloads to perform long-range target acquisition and tracking tasks. Their actual operating environment typically involves a wide temperature range. When performing continuous zoom tasks in complex physical environments with wide temperatures, the control system is required to overcome environmental interference and electromechanical hysteresis to maintain focus matching during the zoom process.

[0003] Currently, the common control strategy for continuous zoom television cameras is a static lookup table zoom tracking algorithm based on discrete sampling, i.e., the lookup table method. This algorithm cyclically collects the discrete position data of the current zoom motor through a serial communication bus and substitutes it into the static zoom-focusing mapping polynomial to calculate the theoretical focus coordinates. When constructing this underlying model architecture, conventional applications usually set the transmission and execution of control commands to an ideal state with no delay, and assume the coupling relationship between zoom and focus to be a static geometric mapping. This strategy shows certain applicability when dealing with low-speed zoom conditions.

[0004] For example, there is a Chinese patent document with publication number CN109525780B, which provides a video-linked camera lens zooming method. By using methods such as lookup tables, machine learning, and neural networks, the permissible range of the field of view corresponding to the current direction is predicted based on historical data. When the current field of view of the camera lens is within this range, the zooming action is not performed.

[0005] However, in continuous zoom operations under wide temperature conditions, the camera control system inherently experiences a system-wide latency from command communication decoding to the execution of the underlying motors. When the zoom lens is in continuous motion, its position is a dynamic variable that changes over time, and the lens continues to undergo physical displacement during this latency period. Traditional static lookup table zoom tracking algorithms, which use a time-delay-free static geometric mapping model and take historical position data as the control reference for the current moment, fail to provide feedforward compensation for the dynamic sliding displacement of the zoom lens during the inherent system latency period. This results in the focusing motor's control command lagging behind the actual physical position of the zoom lens when facing high-speed continuous zoom operations, especially when mechanical damping fluctuates under wide temperature conditions. This causes dynamic phase lag in the camera control system, leading to image defocusing and blurring during zooming. Summary of the Invention

[0006] To address the issue of image defocusing and blurring in continuous zoom television cameras caused by the failure of traditional static lookup table zoom tracking algorithms to promptly compensate for the dynamic sliding displacement of the zoom lens, resulting in the control commands of the focusing motor lagging behind the actual physical position of the zoom lens, this invention proposes a wide-temperature adaptive continuous zoom television camera control method and system.

[0007] On one hand, the present invention provides a wide-temperature adaptive continuous zoom television camera control method, characterized in that it includes: The system acquires the zoom-focus mapping relationship stored inside the continuous zoom television camera, the inherent integrated time delay of the control system, the mechanical inertia time delay of the focusing motor, the depth of focus of the camera lens, and the absolute position data of the zoom motor. The transient motion parameters of the zoom motor are determined based on absolute position data. These transient motion parameters include transient operating speed and transient acceleration. The transient operating speed is multiplied by the inherent comprehensive time delay to obtain a linear displacement prediction term. A nonlinear displacement compensation term is calculated based on the transient acceleration, the inherent comprehensive time delay, and a dynamic damping weight constructed from the transient acceleration and the average acceleration amplitude within a preset historical sampling window. The linear displacement prediction term and the nonlinear displacement compensation term are superimposed to obtain the dynamic lead displacement of the zoom motor. Based on the absolute position data, a reference focus position is mapped in the zoom focus mapping relationship. The superimposed value of the absolute position data and the dynamic lead displacement is substituted into the zoom focus mapping relationship to obtain the predicted focus position. The absolute value of the difference between the reference focus position and the predicted focus position is calculated as the displacement deviation. A rational fraction is constructed with the displacement deviation as the numerator and the sum of the displacement deviation and the depth of focus as the denominator. The function value corresponding to the rational fraction is determined as the defocus vulnerability. The rate of change of the zoom-focus mapping relationship at the absolute position data is extracted as the zoom-focus sensitivity. Based on the zoom-focus sensitivity, transient operating speed, and mechanical inertia delay of the focusing motor, an initial focusing displacement compensation amount is determined. The defocus vulnerability is multiplied by the initial focusing displacement compensation amount to obtain a weighted corrected focusing displacement compensation amount. The weighted corrected focusing displacement compensation amount is superimposed on the predicted focusing position to generate the adaptive feedforward focusing target coordinates of the focusing motor and send them down to drive the focusing motor to perform focusing motion.

[0008] This technical solution starts from the underlying dynamic laws of electromechanical systems and deeply reconstructs the control logic in multiple dimensions: First, by introducing dynamic advance displacement, the sliding deviation of the zoom lens during the delay period of data communication and command parsing is quantitatively predicted, enabling the control model to overcome the inherent time difference of the system and calculate the physical position of the zoom lens in advance at the time of command activation; Second, it overcomes the limitations of single mechanical domain follow-up compensation, and combines the inherent physical properties of the camera optical system, using the deviation between the reference and predicted positions and the optical depth of field to construct the defocus vulnerability. This index, as an adaptive adjustment parameter, can dynamically allocate compensation weights according to the physical optical tolerance of the current focal length, reducing the excessive adjustment of the transmission mechanism in the wide focal depth range while maintaining image clarity, and taking into account both control accuracy and mechanical operation stability; Finally, it integrates the mechanical inertia time delay of the focusing motor for targeted compensation of focusing displacement to cope with the motor start-up lag caused by nonlinear fluctuations in lubrication damping under wide temperature conditions. Through the above mechanism, static geometric mapping is upgraded to dynamic trajectory feedforward prediction with time dimension, which reduces the spatial coupling error between the actual physical position of the zoom lens and the algorithm prediction model at the time of command issuance, alleviates the problem of image defocusing and blurring during continuous zooming, and gives the camera the ability to maintain good focus tracking effect and continuous image clarity in the case of mechanical characteristics fluctuating over time, such as in wide temperature environments.

[0009] Furthermore, transient motion parameters include transient running speed and transient acceleration; the inherent comprehensive delay of the control system is the sum of the serial communication command transmission and reception delay, the command parsing delay of the microcontroller unit inside the continuous zoom TV camera, and the drive response delay of the zoom motor; the acquisition timestamp is acquired synchronously when acquiring the absolute position data of the zoom motor.

[0010] Furthermore, the transient motion parameters of the zoom motor are determined based on the absolute position data, including: establishing a circular buffer queue to store the absolute position data of the zoom motor in the three most recent times, and combining the acquisition timestamps to calculate the original transient operating speed and original transient acceleration of the zoom motor through the discrete backward difference method, so as to constitute the transient motion parameters of the zoom motor.

[0011] Furthermore, the nonlinear displacement compensation term of the zoom motor is determined as follows: a theoretical nonlinear displacement prediction term is calculated based on transient acceleration and inherent integrated time delay; a dynamic damping weight is determined by the relative magnitude of the transient acceleration amplitude and the pre-acquired historical average transient acceleration amplitude; the nonlinear displacement compensation term of the zoom motor is obtained by multiplying the theoretical nonlinear displacement prediction term with the dynamic damping weight.

[0012] This technical solution constructs a noise-resistant and adaptive adjustment mechanism to reduce the interference of high-frequency mechanical vibration on the feedforward system. By extracting the ratio of the current acceleration amplitude to the average amplitude within the historical sliding window and mapping it to a dynamic damping weight, when high-frequency noise caused by mechanical vibration occurs, this ratio is limited by the historical average, and the weight is correspondingly suppressed, thereby reducing spurious compensation. Conversely, when the operational input generates a genuine acceleration command, this ratio increases, and the weight increases accordingly, allowing the system to introduce nonlinear compensation as needed. This mechanism helps the control algorithm distinguish between normal operational input and external environmental noise.

[0013] Furthermore, depth of focus is a physical constant determined by both the aperture number set by the continuous zoom television camera and the pixel size of the internal image sensor.

[0014] Furthermore, the adaptive feedforward focusing target coordinates of the focusing motor are generated and sent out, including: synchronously sending the focusing control command and zoom motion control command, which encapsulate the adaptive feedforward focusing target coordinates, to the continuous zoom television camera within the same serial communication cycle.

[0015] This technical solution includes focusing control commands and zoom motion control commands that include adaptive feedforward focusing target coordinates. These commands are packetized and sent synchronously within the same serial communication cycle, maintaining time consistency of commands at the physical transmission stage. This ensures that the compensated position coordinate pairs can be delivered to the underlying hardware and executed according to a preset phase relationship, thereby reducing mechanical follow-up errors caused by communication.

[0016] On the other hand, the present invention provides a wide-temperature adaptive continuous zoom television camera control system, which includes a control host and a continuous zoom television camera. The control host is bidirectionally connected to the continuous zoom television camera via a serial communication bus. The control host includes a memory and a processor. The memory stores a computer program that can be executed by the processor. When the processor executes the computer program, it implements the steps of any wide-temperature adaptive continuous zoom television camera control method.

[0017] The present invention has the following effects: This invention constructs an adaptive feedforward control architecture. By introducing dynamic advance displacement to compensate for the sliding deviation caused by the comprehensive time delay, and combining the depth of focus to construct a vulnerability index as the adjustment basis, it integrates mechanical inertia time delay for focusing displacement compensation. This transforms static mapping into dynamic trajectory prediction with a time dimension, reducing the spatial error between the actual physical position of the zoom lens and the algorithm prediction model at the time of command issuance. This alleviates the problem of image defocusing and blurring during zooming, and helps the camera maintain good focus tracking effect and image clarity under dynamic conditions. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a comparative schematic diagram showing the change of the absolute error of focus following under high dynamic zoom conditions between the present invention and the traditional method over time. Figure 3 This is a schematic diagram showing the statistical comparison of the average absolute error of the present invention and the traditional method over a wide temperature range. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0020] This invention provides a control method for a wide-temperature adaptive continuous zoom television camera, such as... Figure 1 As shown, it includes: S1: Acquire the transient motion parameters of the continuous zoom television camera.

[0021] This embodiment provides a wide-temperature adaptive continuous zoom television camera, which integrates a main control microprocessor unit (MCU), a zoom motor drive module, a focus motor drive module, and a high-precision absolute position encoder. The zoom motor drive module drives the zoom lens group to move along the optical axis to change the focal length, and the focus motor drive module drives the focus lens group to move to adjust the image sharpness. Before the television camera leaves the factory, a zoom-focus mapping relationship table is calibrated and pre-stored using a precision optical platform; this is typically represented by a high-order polynomial fitting curve. This mapping table defines the theoretical absolute position coordinates that the zoom motor must correspond to in order to ensure image sharpness at each absolute position coordinate. Furthermore, due to the physical limitations of the hardware circuitry, the system has an inherent overall time delay, manifested as the time difference between the issuance of the control command and the activation of the zoom motor actuator.

[0022] Before performing high-precision dynamic feedforward prediction, considering that in actual special optoelectronic load tracking tasks, the serial communication and command parsing between the control host and the camera's underlying driver are non-instantaneous, and the motor actuator has a start-up response process, it is necessary to obtain the transient motion parameters of the continuous zoom television camera for feedforward prediction compensation.

[0023] Next, during the operation of the zoom motor, any time point was used as the sampling time. The zoom motor is read at the sampling time using an absolute position encoder. absolute position data And simultaneously record the system acquisition timestamp at that moment.

[0024] Finally, using the circular buffer queue stored within the control system, the sampling time is extracted. Sampling time and sampling time Three consecutive absolute position data , , The zoom motor at the sampling time was calculated using the discrete backward difference method. transient operating speed With the zoom motor at the sampling time transient acceleration Together, these two constitute the transient motion parameters.

[0025] The extraction of transient motion parameters is based on the classic Taylor series expansion theory in numerical analysis. In discrete digital control systems, the derivatives of continuous differentiable trajectories in physical space cannot be directly obtained from continuous functions. However, through discrete backward difference approximation, the derivatives can be derived when the truncation error is minimal. This allows for the reconstruction of the system's first and second-order dynamic characteristics, namely the characteristics of velocity and acceleration, at a precise scale. In current digital signal processing scenarios, this forms the standard mathematical foundation for achieving non-parametric motion state perception.

[0026] It should be noted that during the initial two sampling periods after the system starts up or is reset, due to insufficient historical position data in the circular buffer queue, the control system sets the initial values ​​of transient running speed and transient acceleration to zero by default, and directly uses the zero values ​​to participate in subsequent compensation calculations.

[0027] In one example, the sampling period of the continuous zoom television camera is set to 0.01 seconds, and the inherent composite delay of the control system is known to be measured. Seconds, ambient temperature Mechanical inertia time delay of the focusing motor at 25 degrees Celsius seconds, depth of focus Millimeters, considering the latency jitter characteristics of bus communication and task scheduling in actual digital control systems, this inherent overall latency... The system can further use internal timers combined with real-time communication timestamps for dynamic measurement updates to replace offline calibration constants.

[0028] At sampling time The system reads the zoom motor at the sampling time. absolute position data Millimeters, absolute position data at the previous sampling time Millimeters, absolute position data at the previous sampling time. Millimeters, zoom motor at sampling time transient operating speed Millimeters per second, zoom motor at sampling time transient acceleration Millimeters per second squared, with the negative sign indicating that the zoom lens group is decelerating.

[0029] After calculating the velocity and acceleration, they need to be smoothed using a first-order low-pass filter algorithm or a moving average filter algorithm to eliminate high-frequency glitches introduced by encoder quantization noise.

[0030] S2: Determine the dynamic advance displacement of the zoom motor.

[0031] Accurately acquiring the zoom lens group at the sampling time After the transient operating characteristics are observed, in order to compensate for the inherent integrated time delay of the control system... The resulting displacement gap must be estimated using feedforward. However, the actual zoom drive system is affected by the nonlinear resistance of the cam groove and the viscosity variation of the lubricating medium under wide temperature conditions. Its acceleration often includes mechanical vibration noise caused by gear meshing. If the full acceleration is directly used for extrapolation prediction, it will lead to high-frequency abnormal fluctuations in the predicted coordinates.

[0032] Therefore, this step aims to establish a displacement prediction mechanism with adaptive suppression capabilities. Based on the transient motion parameters and the inherent integrated time delay of the control system, the dynamic lead displacement of the zoom motor is determined through nonlinear damping modulation.

[0033] The predicted displacement is decoupled into uniform and variable terms, and a nonlinear damping function based on historical acceleration statistics is introduced. The purpose is to suppress noise interference on prediction accuracy when the zoom process is relatively smooth, and to provide sufficient feedforward to align the physical trajectory when the camera operator quickly adjusts the focal length to track the high-speed target and generates real large acceleration, so as to determine the real landing point coordinates of the zoom lens group after the command is issued.

[0034] Specifically, the product of the transient operating speed of the zoom motor at the sampling time and the inherent comprehensive time delay of the control system is determined as the linear displacement prediction term; the ratio of the amplitude of the transient acceleration to the average amplitude of the transient acceleration within the historical sliding window is input into the nonlinear damping mapping function to obtain the dynamic damping weight; the dynamic damping weight is multiplied by the theoretical nonlinear prediction term to obtain the nonlinear displacement compensation term; the linear displacement prediction term and the nonlinear displacement compensation term are superimposed to obtain the linear displacement prediction term and the linear displacement compensation term at the sampling time. The dynamic advance displacement.

[0035] The following calculation relationship must be satisfied:

[0036] In the relation: For the zoom motor at the sampling time The dynamic advance displacement. For the zoom motor at the sampling time transient operating speed, For the zoom motor at the sampling time transient acceleration, Due to the inherent overall time delay of the control system, The average amplitude of transient acceleration of the zoom motor within a preset historical sampling window can be dynamically updated using a first-order exponential moving average algorithm or a first-order low-pass filter. As a preset constant for preventing misjudgment of minute accelerations in the system, even when the zoom motor is running at a constant speed or stationary, the minimum scale accuracy of its position sensor (i.e., the absolute position encoder) will cause extremely minute digital jumps in the read values. After differential calculation, this normal jump will produce a false minute acceleration. This is set to a fixed value slightly larger than the spurious small acceleration, to ensure that the denominator does not tend to zero in a steady state, thus enabling the damping weight to correctly identify and suppress underlying noise interference. It is the absolute value symbol.

[0037] It should be noted that, Another method for obtaining the data is to construct a fixed-length FIFO (First-In-First-Out) circular data queue in the control system's memory as a historical sampling window, and set its sample count. The sampling points are 50 discrete points. The specific selection principle for this value is to cover the amount of data for one complete low-frequency oscillation cycle of the mechanical system in order to filter out the interference of high-frequency mechanical noise.

[0038] In each new sampling period, the latest calculated absolute value of transient acceleration is pushed into the queue, and the oldest data is removed. The arithmetic mean of the data in the queue is then calculated to achieve... Sliding updates; The specific method for obtaining this information is as follows: During the camera's factory testing phase, the absolute value of the maximum fluctuation amplitude of the filtered acceleration is collected under both stationary and constant low-speed operation conditions of the zoom motor as the peak noise floor. and will Setting it to 1.5 times the peak noise level ensures that the denominator of the damping term does not approach zero under any operating condition, thus avoiding calculation overflow, and also precisely sets the threshold for distinguishing effective instructions from underlying background noise.

[0039] In this relation, A dynamic damping weight is constructed using a first-order rational fraction to reflect the deviation of the current transient acceleration from historical background noise. The larger this value, the more likely the currently detected acceleration is a real acceleration caused by a large-scale, highly dynamic sudden zoom operation, far exceeding the historical background noise. The more likely the currently detected acceleration change is caused by real operation commands rather than environmental noise, the more the system will release a larger compensation ratio to ensure that the focusing system can closely follow the violent acceleration and deceleration of the zoom lens. Conversely, if the currently detected acceleration is close to the historical average, it is likely due to minor mechanical vibrations caused by gear meshing gaps inside the TV camera or external wind loads. The system will suppress this unreliable compensation calculation to prevent low-level interference signals from causing image oscillations.

[0040] In this relation, This constitutes a linear displacement prediction term, which characterizes the displacement increment based on the current uniform velocity assumption. It is the main reference for prediction. The larger this term is, the longer the uniform sliding distance that the zoom lens will produce based on the current speed within the communication delay. When the camera is in a stable target zooming state, this term dominates.

[0041] In this relation, This constitutes a theoretical nonlinear prediction term, used to calculate the theoretical displacement increment based on the uniformly accelerated linear motion model. The larger this term is, the more violent the acceleration and deceleration of the zoom lens, and the more obvious the nonlinear displacement deviation generated within the time delay.

[0042] The entire relation is constructed using a first-order rational fraction to create a numerically adaptive damping buffer layer. By determining the dynamic damping weight, the nonlinear displacement part in the uniformly accelerated linear motion model is finely adjusted by multiplicative damping control, which conforms to the prediction of Newtonian classical kinematics and the characteristics of mechanical damping.

[0043] When a television camera experiences minor mechanical vibrations due to internal gear meshing clearances or external wind loads... Compared with historical average When the ratio approaches half, the damping weight is small, and the system suppresses this unreliable nonlinear compensation, ensuring the smoothness of the motor output signal and preventing image oscillation. When the camera control system receives a sudden zoom command with a wide range and high dynamics, and the zoom motor generates a real acceleration several times greater than the background noise, the ratio in this fraction increases sharply, the damping weight quickly approaches 1, the nonlinear displacement compensation term is fully activated, and the system outputs full compensation, ensuring that the focusing motor can follow closely.

[0044] The dynamic advance displacement reflects the predicted sliding deviation of the zoom lens during the delay period between data communication and command parsing (i.e., the inherent comprehensive system delay). It is used to fill displacement gaps across system time differences. A larger value indicates that the zoom motor is currently operating at extremely high speed or experiencing very rapid acceleration or deceleration. In the tens of milliseconds between the control system sending a command and the actuator generating an action, the zoom lens travels a longer physical distance due to inertia and speed, requiring a larger advance displacement margin from the system. Conversely, a smaller value indicates that the zoom motor is in a slow creeping or stationary state, with almost no additional physical sliding of the lens during the command delay period, requiring less advance correction.

[0045] Continuing from S1: millimeters per second millimeters per second squared Seconds, set the average amplitude of transient acceleration within the historical sliding window counted by the controller. millimeters per second squared .

[0046] Linear displacement prediction term: millimeters; Dynamic damping weight: ; Nonlinear displacement compensation term: millimeters; It should be further explained that the displacement or position data of the zoom motor acquired and calculated in this invention, in its physical essence, refers to the actual physical displacement generated by the zoom lens group driven by the zoom motor through the transmission mechanism along the optical axis. The two are strictly one-to-one corresponded by a fixed mechanical transmission ratio. In order to fit the processing logic of the underlying control system for acquiring encoder data, it is referred to as the displacement data or absolute position data of the zoom motor in this invention, and will not be elaborated further.

[0047] Therefore, the zoom motor at the sampling time is obtained. Dynamic advance displacement Millimeters.

[0048] S3: Predict the focus position and calculate the defocus vulnerability.

[0049] After calculating the predicted physical coordinates of the zoom lens group after the signal transmission delay period, directly and indiscriminately driving the focusing motor to perform large-step following would lead to unnecessary frequent adjustments in certain optically insensitive focal lengths. Therefore, this step derives the predicted focusing position of the focusing motor at future moments based on the dynamic lead displacement, and determines the defocus vulnerability in conjunction with the depth of focus.

[0050] Depth of focus (DFO) refers to the maximum permissible positional deviation of the focusing lens group along the optical axis under the constraints of current aperture and pixel size, in order to maintain a sharp image. In other words, it's the maximum physical distance the focusing lens group is allowed to move back and forth on a mechanical guide rail while maintaining image sharpness on the optical sensor at the current aperture and focal length. It's essentially a safety tolerance zone for mechanical movement; as long as the actual physical positional deviation of the focusing lens falls within this distance threshold, the captured image will not exhibit blurriness visible to the human eye or the algorithm.

[0051] The zoom focusing mechanical mapping characteristics of a television camera determine that it has different follow-up error amplification rates at different focal lengths. At the wide-angle end, the mapping curve has a gentle slope, and the predicted focus deviation caused by slight zoom sliding is minimal, falling entirely within the fixed depth-of-focus tolerance margin, and will not cause image blurring. However, at the telephoto end, the mapping curve is extremely steep, and even a tiny zoom sliding deviation will be amplified into a huge focus displacement deviation, causing the image to deviate from the fixed depth-of-focus range.

[0052] Therefore, this step aims to establish a flexible adjustment mechanism based on physical optical characteristics. By calculating the deviation of the predicted position relative to the reference position and mapping it to the depth of focus, a dimensionless vulnerability index is generated. The purpose is to allow the control system to sense the degree of defocusing in the current optical state, automatically increasing the feedforward intensity in stages with a high risk of image quality degradation, and suppressing the mechanical adjustment amplitude when the depth of focus is sufficient.

[0053] Specifically, based on the absolute position data of the zoom motor at the sampling time, the zoom focus mapping relationship is... The reference focusing position is obtained by mapping, and the absolute position data is combined with the dynamic advance displacement. By superimposing the sub-subjects into the mapping relationship, the predicted focus position is obtained. The absolute value of the deviation between the two is calculated, and the ratio of this deviation to the depth of focus is used as the independent variable of the Softsign function to determine the defocus vulnerability of the zoom motor at the sampling time.

[0054] The following calculation relationship must be satisfied:

[0055] In the relation, For the zoom motor at the sampling time The fragility of defocusing For the zoom motor at the sampling time absolute position data The reference focusing position at that time The dynamic advance displacement was added to the prediction of the zoom motor. The corresponding predicted focus position at that time For focal depth, It is the absolute value symbol.

[0056] This relation constructs a Softsign function, which is a smooth, non-linear function. The standard definition of the Softsign function is:

[0057] This place makes After substituting and simplifying, we obtain the relation. This reflects the predicted displacement deviation. As the dependent variable, is the independent variable.

[0058] Based on the characteristics of the Softsign function, when the camera is in wide-angle scanning mode, the predicted displacement deviation is extremely small because the zoom focus mapping curve has a gentle slope. Approaching 0, the output of the Softsign function is extremely small, close to 0. This is equivalent to informing the motor control system that the current optical imaging is very safe, eliminating the need for frequent adjustments to the focusing motor due to minor time delay errors, thus reducing mechanical wear. When the camera zooms out to lock onto the target, the zoom-focus mapping curve has a steep slope relative to a fixed depth of focus. Even a slight zoom delay can cause a sharp increase in the absolute value of the predicted displacement deviation, making... The nonlinearity increases sharply, and the saturation characteristic of the Softsign function makes the output result nonlinearly approach 1. The system determines that it is currently in a vulnerable area that is prone to defocusing, and a correction intervention is needed to prevent the target from becoming blurred.

[0059] As can be seen, defocus vulnerability reflects the degree to which the current optical state is prone to defocusing, i.e., the probability of image blurring. It characterizes the proportion of displacement deviation caused by communication prediction to the optical tolerance boundary, projecting geometric error into optical tolerance. A larger value, closer to 1, means the camera is in a long focal length target-locking state, and the zoom-focus mapping curve is extremely steep. At this point, even a small zoom delay will cause a sharp increase in prediction displacement deviation, and a very small error will cause the image to deviate from the fixed depth of focus range. The system determines that the current defocus risk has increased and needs to apply the maximum proportion of correction intervention, fully introducing mechanical compensation to prevent target blurring. Conversely, a smaller value, closer to 0, means the camera is in a wide-angle survey state, the mapping curve slope is gentle, and the prediction focus deviation caused by slight zoom slippage is minimal, falling entirely within the fixed depth of focus tolerance margin. The system determines that the current optical imaging is very safe and will actively suppress the compensation force on the focusing motor to reduce over-adjustment and mechanical wear.

[0060] This relation is constructed based on the theories of circle of confusion and depth of field tolerance in optical engineering, and combined with the Softsign function from the field of control. It projects absolute geometric spatial errors into a relative tolerance space dominated by lens optical parameters, and utilizes the computational structure of pure algebraic division and the asymptote convergence property of the Softsign function to smoothly map unbounded physical distance errors into a bounded one. A probabilistic evaluation space, reflecting the risk of image blurring, is established, enabling the transformation from the geometric-physical domain to the control evaluation domain.

[0061] Continuing from S2: millimeters millimeters Millimeters, assuming zoom focus mapping relationship Locally, it simplifies to a linear function within this interval. Reference focus position Millimeters, predicted focal position Millimeters, absolute value of deviation Millimeters, zoom motor at sampling time Defocus fragility .

[0062] S4: Extract zoom focus sensitivity and determine focus displacement compensation amount.

[0063] In extreme physical environments such as wide temperature ranges, the thermal expansion and contraction of optical lenses and the drift of electronic component parameters can have a certain impact on image sharpness. However, compared to the nonlinear mechanical damping abrupt change caused by the drastic temperature-induced viscosity change of the lubricating medium in the mechanical guide rail, the former's impact exhibits a long-period, gradual change characteristic and can be compensated for through mature methods such as factory temperature calibration and periodic online calibration. It is a secondary factor affecting the instantaneous sharpness of the image in high dynamic continuous zoom scenarios.

[0064] Meanwhile, the continuous zoom control system targeted by this invention adopts a master-slave follower architecture. The zoom motor, as the active axis, directly responds to external zoom control commands. Its damping changes over a wide temperature range only affect the execution timeliness of the zoom action. Ultimately, it can still achieve accurate arrival of the target position through position closed loop, without causing instantaneous defocusing of the image during high dynamic zoom. On the other hand, the focus motor, as the driven axis, is completely subject to the real-time physical position of the zoom axis and must maintain strict phase synchronization with the dynamic movement of the zoom axis.

[0065] Therefore, under wide temperature conditions, the underlying response lag of the focusing motor caused by the nonlinear change in the viscosity of the lubricating medium is the core reason for the instantaneous defocusing of the image during high dynamic continuous zoom.

[0066] After completing the global trajectory prediction and optical adjustment weight allocation, microscopic corrections are also needed to address the mechanical starting characteristics of the focusing axis itself. Considering the inherent rotor physical inertia of the focusing motor and its transmission mechanism, and the significant nonlinear abrupt change in the viscous resistance of the internal grease under wide temperature conditions, especially low temperature conditions, during the high dynamic tracking process of continuous zoom, when the slope of the mapping curve increases and the focusing motor is required to produce a step-like speed change, there is an objective dynamic response hysteresis period of tens of milliseconds from the time the motor receives the acceleration command to the time it actually outputs the matching speed, i.e., mechanical inertia delay.

[0067] During this extremely brief pause window, the zoom lens assembly does not stop; it continues to generate physical displacement due to inertia. Furthermore, the linkage rate between the zoom trajectory and the focus trajectory is not a proportional linear relationship; its instantaneous linkage strength depends entirely on the slope of the tangent line of the higher-order cam curve at the current coordinate.

[0068] Therefore, this step aims to establish a local dynamic compensation mechanism for the focusing driven axis. By extracting the rate of change of the zoom focusing mapping relationship as the zoom focusing sensitivity, and combining it with transient motion parameters, the focusing displacement compensation amount during the zoom process is determined.

[0069] The first-order difference rate of change of the mapping curve at the current discrete coordinates is extracted as the spatial geometric sensitivity. The additional error caused by the stagnation of the dynamic response is calculated and converted into the displacement step compensation increment required for the focusing axis, thereby aligning the mechanical action phase.

[0070] Specifically, the rate of change of the zoom-focus mapping relationship at the absolute position data of the zoom motor is extracted as the zoom-focus sensitivity. Specifically, it is determined by the zoom-focus mapping relationship. The result is obtained by dividing the position difference between two consecutive sampling points by the position difference of the zoom motor with zero dead zone constraint; based on zoom focus sensitivity. Transient operating speed of zoom motor And the mechanical inertial time delay of the focusing motor subject to temperature correction. The product determines the zoom motor at the sampling time. Focused displacement compensation amount .

[0071] It should be noted that when zooming focuses, the mapping relationship... When stored in discrete data table format, the sensitivity is extracted using the following difference relationship formula; if If stored in the form of a polynomial function, its first-order analytic derivative is directly calculated by the MCU as the zoom focus sensitivity.

[0072] The sensitivity of the difference relation extraction satisfies the following form:

[0073] In the relation, Let be the zoom focusing sensitivity of the zoom motor at sampling time t. and These represent the theoretical focusing positions corresponding to two adjacent sampling times. For symbolic functions, To find the maximum value function, To prevent division by zero overflow during calculation, the parameter values ​​are obtained by directly reading the theoretical maximum resolution of the zoom absolute position encoder, denoted as: Millimeters are used to ensure that hardware-level physical limits are not exceeded, and || is the absolute value symbol.

[0074] In the specific implementation, to prevent the underlying microcontroller from crashing due to floating-point division by zero overflow when the motor is stationary or operating at extremely low speeds, an overflow protection mechanism is introduced. Specifically, when the independent variable of the sgn() function is 0, i.e., the zoom motor is completely stationary and the position difference is 0, its function value is forcibly set to 1. If this forced setting is not made, the natural mathematical value of sgn(0) is 0, which will cause the denominator of the entire formula to be multiplied to zero again, making the division-by-zero prevention mechanism ineffective. Forcing it to 1 and combining it with the maximum value function max() can ensure that the absolute value of the denominator is always the lower limit threshold under any extreme dwell conditions. This eliminates the risk of system crashes due to division by zero in the underlying operations while maintaining the continuity of the mathematical model.

[0075] In the differential relationship for extracting sensitivity, the numerator represents the difference between the reference focusing positions corresponding to two adjacent sampling times, i.e., the theoretical increment of focusing motion; in the denominator, the max function is used in combination with... This is to force a safety lower limit on the denominator when the zoom motor is stationary or undergoing extreme low-speed creep that causes the position change to approach zero. However, simple absolute value operations and max function filtering inevitably strip away the sign of the zoom displacement, that is, they lose the physical direction polarity of the zoom lens's actual movement and cannot distinguish whether it is pushing away or pulling in.

[0076] Therefore, the core purpose of using the sgn sign function is to reassign the original directional polarity of the zoom motion after the minimum safety filtering. Preserving this polarity is crucial for the normal execution of subsequent servo compensation. If the sgn function is removed, the extracted sensitivity will lose its directional attribute and remain permanently positive. This will result in the loss of reverse adjustment capability when calculating the focus displacement compensation amount. When the actual physical optical path requires the focusing lens to move in the reverse direction to maintain the conjugate plane, the command lacking polarity will drive the motor to continue moving in the forward direction, which will not only fail to compensate but will also amplify the defocusing error many times over.

[0077] By restoring the polarity of the sgn function, the multiplication of the two precisely reconstructs the directional zoom displacement increment with hardware overflow protection. Thus, by dividing the numerator and denominator, the system can stably and accurately extract the slope of the local spatial tangent with the correct directional attribute at the current sampling position—that is, how many units the focusing lens needs to move for each unit of zoom lens movement—ensuring that the subsequently generated displacement step compensation increment is absolutely correct in both physical amplitude and direction of motion.

[0078] Zoom focus sensitivity reflects the rate of change of the local spatial geometry of the zoom focus mapping relationship at the current discrete coordinates. Essentially, it's the slope of the tangent at the current position, representing the proportional relationship between the zoom lens's movement and the focusing lens's movement. A higher value indicates extremely sensitive zoom and focus linkage at the current focal length; any tiny physical displacement of the zoom lens will be amplified by the high slope, requiring the focusing motor to produce a large-scale compensation. This typically occurs at the telephoto end. Conversely, a lower value means lower zoom and focus linkage sensitivity; even if the zoom lens moves a distance, the focusing lens only needs minimal or no movement to maintain image sharpness. This typically occurs at the wide-angle end.

[0079] The focused displacement compensation amount satisfies the following relationship:

[0080] In the relation, For the zoom motor at the sampling time The amount of focused displacement compensation. The zoom focus mapping relationship at the sampling time zoom focus sensitivity, This refers to the transient operating speed of the zoom motor. To focus the motor on ambient temperature Mechanical inertial time delay.

[0081] It should be noted that, Considering that the viscosity of the lubricating grease inside the focusing guide rail changes non-linearly under a wide temperature range (from -40℃ to +60℃), thus altering the dynamic viscous resistance and response delay of the motor, the system pre-calibrates and stores a two-dimensional lookup table of temperature and response delay before leaving the factory. During operation, the system reads the temperature from the camera's internal temperature sensor at the sampling time. ambient temperature The mechanical inertial time delay corresponding to the dynamic temperature is obtained by interpolation from a table. .

[0082] In the relation, This reflects the actual physical distance the zoom lens assembly travels forward within the dynamic hysteresis window, when the focusing motor cannot accelerate instantaneously due to mechanical inertia. This is compared with sensitivity. Multiplication and fusion: If the camera operates in a low-temperature environment, the lubricating grease may solidify, leading to... The multiplication factor amplifies the effect directly; if this happens to be at the focal length where the cam slope is extremely high... The larger the value, the more any tiny tracking delay will be amplified into a severe image plane deviation. This relationship accurately quantifies the feedforward coordinate increment that the focus axis must immediately replenish in order to catch up with the gap created by the zoom motor due to response inertia during high-dynamic-range linkage. Conversely, if operating in normal or high-temperature environments, the grease has good fluidity. Extremely small, and the camera is positioned precisely at the wide-angle end where the zoom curve is relatively flat. The difference is extremely small. At this point, any tiny difference in physical displacement will not cause significant defocusing. The compensation coordinate increment calculated by the product will also automatically shrink to a tiny level, thereby avoiding unnecessary high-frequency oscillations in the motor.

[0083] The sensitivity relationship for extraction is constructed based on Newton's difference quotient limit theory in calculus, combined with the anti-saturation constraint model in nonlinear control theory. Traditional differentiation can lead to floating-point overflow and system crashes when the zoom motor is stationary or experiences extremely minor vibrations. By introducing... Functions with directional constraints and absolute value threshold lower bounds The hardware overflow protection mechanism ensures the computational stability and robustness of the underlying control algorithm under extremely low-speed conditions.

[0084] In summary, the focus displacement compensation reflects the actual physical distance the zoom lens assembly travels forward within the dynamic hysteresis window—the window during which the focus motor cannot instantly accelerate due to mechanical inertia. The larger the value, the more likely the camera is operating in a low-temperature environment (due to solidified grease causing significant mechanical response delay) and is at a high-magnification telephoto end. This requires the system to immediately output a large step coordinate increment to compensate for the physical gap created by the slow start-up of the underlying motor and the zoom lens. Conversely, a smaller value indicates the camera is operating in a high-temperature environment (due to good grease flow and minimal inertia) or at a flat wide-angle end. In this case, the system determines the focus motor's following pressure is extremely low, and the compensation increment automatically shrinks to a tiny level, avoiding high-frequency oscillations during high-speed zooming.

[0085] Continuing from the aforementioned data: millimeters millimeters millimeters per second, at room temperature Seconds, set the encoder resolution minimum value Millimeters, known simplified mapping relationship millimeters Millimeters.

[0086] Location difference millimeters, greater than ; Zoom focus sensitivity ; zoom motor at sampling time Focused displacement compensation amount Millimeters.

[0087] S5: Generate and distribute adaptive feedforward focusing target coordinates.

[0088] After calculating the main trajectory prediction coordinates representing global communication prediction, the defocus vulnerability representing optical flexible adjustment, and the focusing displacement compensation amount representing the inertial compensation gap at the actuator bottom layer, this step aims to complete the final integration from the control algorithm logic layer to the physical communication execution layer.

[0089] To avoid imposing redundant compensation amounts, which are permissible under optical conditions, on the focusing motor and thus disrupting the stable state, this step uses defocus vulnerability correction to weighted correction of the focusing displacement compensation amount, reconstructs and sends out the final adaptive feedforward focusing target coordinates.

[0090] By using a vulnerability index as a nonlinear gain, the mechanical compensation is strictly constrained and ultimately superimposed onto the main prediction baseline to synthesize a final coordinate system that encompasses both prediction and compensation attributes. Subsequently, data frame encapsulation intervention is performed at the underlying physical link protocol to distribute this coordinate system.

[0091] Specifically, the defocus vulnerability of the zoom motor at the sampling time is weighted and multiplied by the focusing displacement compensation amount, and the product is then superimposed on the predicted focusing position derived from the advance displacement. Above, the zoom motor is generated at the sampling time. Adaptive feedforward focusing target coordinates This coordinate represents the theoretically optimal physical target point that the focusing lens should accurately reach at the moment the future command takes effect, after offsetting control communication delays, taking into account optical depth tolerance, and compensating for the mechanical hysteresis of motor dynamic tracking.

[0092] Specifically, the following relationship is satisfied:

[0093] In the relation, To focus the target coordinates of the motor at sampling time t using adaptive feedforward focusing, reflecting the final reconstructed hysteresis-free electromechanical servo command, Some of these are globally predicted focus locations. For defocusing fragility; This is the amount of focused displacement compensation for local mechanical inertia response.

[0094] The formula superimposes the global macroscopic system communication delay prediction term with the microscopic focus displacement compensation: if the camera rapidly zooms in on the target at the telephoto end, the optical tolerance is compressed. When the value approaches 1, the system fully applies the focus compensation amount, thereby outputting the maximum compensation gain to compensate for communication and start-up delay gaps, ensuring precise focus synchronization; if in a wide-angle slow scan state, As the coordinates approach zero, mechanical compensation is dynamically suppressed, and the final output coordinates are almost identical to the main prediction coordinates, thus avoiding minor high-frequency jitter in the image.

[0095] The construction of this synthetic relation is based on the feedforward-feedback composite control theory and the multi-source nonlinear weighted fusion theory in modern control theory. The first term of the relation establishes the deterministic feedforward channel backbone based on the system identification model, namely the zoom mapping curve and the time delay; while the second term introduces an adaptive dynamic compensation increment based on environmental state space observation, namely temperature, optical limitations, and differential sensitivity. The superposition of the two realizes the reconstruction of the hysteresis errors of multiple independent time domains in a single spatial dimension.

[0096] It should be noted that the adaptive feedforward focusing target coordinates are mathematically represented as scalar values ​​in a single dimension, such as the absolute pulse count of the encoder or millimeter displacement, rather than vector coordinates in two- or three-dimensional space. This is determined by the inherent internal opto-mechanical-electrical physical architecture and servo control principles of continuous zoom television cameras. Specifically, in the mechanical structure of a continuous zoom television camera, both the zoom lens group and the focusing lens group are physically constrained on a highly precise linear optical guide rail. The physical motion mode of the lens is a strict one-dimensional linear reciprocating translation along the optical axis, that is, it has only a single degree of linear motion freedom. In industrial servo motor control and absolute position encoder measurement systems, for such single-degree-of-freedom linear motion mechanisms, the system uses the mechanical limit zero point of the guide rail or the origin of the grating ruler as the absolute reference point. Therefore, by using a monotonically changing absolute distance value, the real-time position of the lens on the guide rail can be uniquely and absolutely located in physical space.

[0097] The data flow following the preceding steps: Predicted focal position millimeters 0.948, the overall focusing displacement compensation is Millimeters, adaptive feedforward focusing target coordinates Millimeters, rounded to three decimal places.

[0098] In summary, the adaptive feedforward focusing target coordinates reflect the theoretically optimal target point that the focusing lens should accurately reach at the moment the future command takes effect, after the control algorithm has compensated for communication delays, evaluated optical depth-of-focus tolerances, and filled the physical lag of motor start-up. A larger dynamic increment in the adaptive feedforward focusing target coordinates means that, under complex and challenging operating conditions such as wide temperature variations, high-speed zoom, and high-magnification long focal lengths, the system has performed significant feedforward prediction and lag intervention, resulting in a greater difference between the issued coordinates and the static lookup table coordinates. Conversely, a smaller dynamic increment in the adaptive feedforward focusing target coordinates means that the system is currently in a stable or wide-angle optical state, requiring no excessive correction. Mechanical compensation is suppressed by dynamic limiting indicators, and the final issued coordinates are almost identical to the main predicted coordinates, ensuring system flexibility and stability.

[0099] It should be noted that in generating the final adaptive feedforward focusing target coordinates Subsequently, to prevent the target coordinates from exceeding the physical travel boundary of the focusing lens due to the superposition of high dynamic compensation, the control host performs a check on the target coordinates before issuing the command. Perform a mechanical soft limit threshold cutoff operation to forcibly limit its value to a preset minimum and maximum effective stroke range. This design avoids the bottom motor from impacting the physical limiter under extreme operating conditions.

[0100] S6: Achieve continuous zoom based on the coordinates of the focusing target using adaptive feedforward.

[0101] After completing the underlying algorithm calculations, the control host encapsulates the target coordinate value of 22.592 mm into a serial hexadecimal data frame instruction packet and sends it synchronously to the underlying driver module within the same communication cycle as the zoom motion control command. This step returns to the hardware execution terminal of the continuous zoom television camera control system, which generates motion control commands based on the adaptive feedforward focusing target coordinates, driving the zoom motor and focus motor to perform high-precision electromechanical servo control to achieve continuous zoom.

[0102] By coordinating the response of the driver's electronic circuits and electromechanical components, the pre-generated target instructions, which include time feedforward and spatial compensation, are strictly executed to complete the closed loop of physical motion.

[0103] Specifically, the low-level microcontroller unit (MCU) inside the camera receives the coordinates of the adaptive feedforward focusing target. After receiving the communication packet containing the zoom motion control command issued synchronously, to avoid sudden changes in motor current and mechanical shock caused by step target commands with advance compensation, the MCU first inputs the absolute coordinate value to an internal trajectory generator, such as a high-frequency position micro-interpolator or spline interpolation algorithm, to generate a smooth transition trajectory. Using the current system control cycle as the beat, the MCU discretizes the instantaneous reference coordinates at each moment. Simultaneously, the MCU reads the current actual physical position pulse count of the focus axis absolute position encoder in real time through its internal high-speed serial peripheral interface bus. After converting the instantaneous reference coordinates output by the trajectory generator into the corresponding theoretical pulse count, the difference between this theoretical pulse count and the actual physical position pulse count fed back by the encoder is calculated in real time to determine the instantaneous position deviation.

[0104] Next, the instantaneous position deviation is sent to the position-velocity dual closed-loop PID (proportional-integral-derivative) controller to calculate the required output control quantity. This control quantity is then converted into a PWM (pulse width modulation) signal with a corresponding duty cycle and injected into the control pin of the H-bridge motor driver chip, such as the L298N or an advanced motor driver IC. The H-bridge circuit controls the actual current flowing into the multiphase winding of the focusing motor according to the PWM signal, thereby generating electromagnetic torque. This drives the focusing lens group to smoothly and quickly move to the corrected target coordinates with high dynamic response characteristics, accurately stopping on the optical conjugate surface. Meanwhile, the zoom motor performs independent closed-loop drive according to the synchronously issued zoom motion control command.

[0105] This complete electromechanical control execution process realizes the feedforward prediction and compensation requirements given by the front-end algorithm, thereby solving the problems of defocusing and blurring during zoom tracking in high-speed target movement or extreme environments.

[0106] To verify the effectiveness of the wide-temperature adaptive continuous zoom control method of this invention, a television camera equipped with a telephoto lens was selected and subjected to actual bench testing under high-frequency operating conditions that fully covered smooth mechanical acceleration, constant speed operation, and high-dynamic rapid deceleration, as well as three typical ambient temperatures: normal temperature, extremely cold temperature, and high temperature. By retrieving the focusing absolute encoder pulse deviation recorded in real time by the bottom-end microcontroller unit (MCU), the following electromechanical servo effect comparison chart was extracted and plotted: like Figure 2 As shown, the transient curve of the absolute focus tracking error over time is presented under high dynamic continuous zoom conditions. Observing the numerical trend, in... Instant In the initial stage of startup, the control effect of this invention and traditional methods is basically the same. However, when entering... Instant During rapid deceleration and high-risk defocusing zones lasting only seconds, traditional methods, due to the lack of compensation for the overall time delay and mechanical lag of the control system, result in phase lag and overshoot caused by the mechanical inertia, with an error of approximately [missing information]. Pulse, exceeded At the absolute critical point of the lens's optical depth of focus, the image experiences severe physical defocusing. In contrast, the adaptive feedforward compensation strategy of this invention uses transient acceleration and nonlinear damping weights to predict the dynamic lead displacement, and dynamically releases the compensation weights based on the defocus vulnerability, suppressing the instantaneous overshoot peak to approximately... The pulse, located below the focus depth safety line, reflects the servo intervention process in which the system continuously performs high-frequency current fine-tuning based on dynamic feedforward coordinates, thus eliminating the risk of macroscopic time delay defocusing.

[0107] like Figure 3 As shown, at room temperature Under the benchmark, the average absolute error of the traditional method throughout the entire process is: Pulse. However, when the ambient temperature suddenly drops... At this time, the viscosity of the guide rail grease surges, causing the mechanical damping to increase non-linearly. Traditional methods, lacking an environmental awareness closed loop, have their average error amplified. Pulse. This invention, by incorporating the temperature-corrected mechanical inertia delay of the focusing motor into the calculation of the focusing displacement compensation, endows the control system with low-level sensing capabilities. This invention is effective at low temperatures. The average error is only Pulse, compared to normal temperature The pulse has only an extremely small amplification, in The error also stabilized at high temperatures. The pulse, a strategy of this invention that integrates defocus vulnerability with temperature hysteresis parameters, isolates the nonlinear interference of a wide-temperature physical environment on mechanical transmission performance and improves the consistency accuracy of focus following.

[0108] In addition, the present invention provides a wide-temperature adaptive continuous zoom television camera control system, which includes a control host and a continuous zoom television camera. The control host is bidirectionally connected to the continuous zoom television camera via a serial communication bus. The control host includes a memory and a processor. The memory stores a computer program that can be run by the processor. When the processor executes the computer program, it implements the wide-temperature adaptive continuous zoom television camera control method in steps S1 to S6.

[0109] The control system reads the zoom encoder values ​​at high frequency, calculates transient motion parameters by the differential algorithm unit inside the processor, and calls the zoom focus mapping polynomial coefficients in the memory. In each control cycle, it completes the above-mentioned nonlinear damping prediction, optical vulnerability assessment and mechanical hysteresis compensation calculation, and finally sends synchronous motion commands to the execution end through the control bus.

[0110] In summary, this invention reconstructs the control logic for focusing by modeling the transient dynamics of the zoom lens group and combining the inherent time delay of the system with the optical depth of focus constraint. This effectively solves the defocusing problem caused by communication lag and mechanical inertia during continuous zooming, ensuring image clarity under wide temperature conditions and high dynamic zooming.

Claims

1. A control method for a wide-temperature adaptive continuous zoom television camera, characterized in that, include: The system acquires the zoom-focus mapping relationship, the inherent integrated time delay of the control system, the mechanical inertia time delay of the focusing motor, the focal depth of the camera lens, and the absolute position data of the zoom motor, all pre-stored inside the continuous zoom television camera. Specifically, it obtains the dynamic mechanical inertia time delay at that temperature by reading the ambient temperature at the sampling time from the camera's built-in temperature sensor and interpolating the data from a table. The inherent overall latency of the control system is the sum of the serial communication command transmission and reception latency, the command parsing latency of the microcontroller unit inside the continuous zoom TV camera, and the drive response latency of the zoom motor; the acquisition timestamp is acquired synchronously when acquiring the absolute position data of the zoom motor; The transient motion parameters of the zoom motor are determined based on absolute position data. The transient motion parameters include transient running speed and transient acceleration. The transient running speed is multiplied by the inherent comprehensive time delay to obtain the linear displacement prediction term. Based on transient acceleration, inherent integrated time delay, and a dynamic damping weight constructed from the transient acceleration and the average acceleration amplitude within a preset historical sampling window, a nonlinear displacement compensation term is calculated, including: calculating a theoretical nonlinear displacement prediction term based on transient acceleration and inherent integrated time delay; determining a dynamic damping weight by the relative magnitude of the transient acceleration amplitude and the pre-acquired average amplitude of historical transient acceleration; and multiplying the theoretical nonlinear displacement prediction term by the dynamic damping weight to obtain the nonlinear displacement compensation term for the zoom motor. The linear displacement prediction term and the nonlinear displacement compensation term are superimposed to obtain the dynamic advance displacement of the zoom motor; based on the absolute position data, the reference focus position is mapped in the zoom focus mapping relationship; the superposition value of the absolute position data and the dynamic advance displacement is substituted into the zoom focus mapping relationship to obtain the predicted focus position. The absolute value of the difference between the reference focus position and the predicted focus position is calculated as the displacement deviation. A rational fraction is constructed with the displacement deviation as the numerator and the sum of the displacement deviation and the depth of focus as the denominator. The function value corresponding to this rational fraction is determined as the defocus vulnerability. The rate of change of the zoom-focus mapping relationship at the absolute position data is extracted as the zoom-focus sensitivity. Based on the zoom-focus sensitivity, transient operating speed, and mechanical inertia delay of the focusing motor, the initial focusing displacement compensation is determined. The defocus vulnerability is multiplied by the initial focusing displacement compensation to obtain the weighted corrected focusing displacement compensation. The weighted corrected focusing displacement compensation is superimposed on the predicted focusing position to generate the adaptive feedforward focusing target coordinates of the focusing motor and send them down to drive the focusing motor to perform focusing motion.

2. The wide-temperature adaptive continuous zoom television camera control method according to claim 1, characterized in that, The transient motion parameters of the zoom motor are determined based on absolute position data, including: A circular buffer queue is established to store the absolute position data of the zoom motor in the three most recent times. Combined with the acquisition timestamp, the original transient operating speed and original transient acceleration of the zoom motor are calculated by the discrete backward difference method to form the transient motion parameters of the zoom motor.

3. The wide-temperature adaptive continuous zoom television camera control method according to claim 1, characterized in that, Depth of focus is a physical constant determined by both the aperture number set in a continuous zoom television camera and the pixel size of its internal image sensor.

4. The wide-temperature adaptive continuous zoom television camera control method according to claim 1, characterized in that, Generate and send out the adaptive feedforward focusing target coordinates of the focusing motor, including: The focus control command and zoom motion control command, which encapsulate the coordinates of the adaptive feedforward focus target, are synchronously sent to the continuous zoom television camera within the same serial communication cycle.

5. A wide-temperature adaptive continuous zoom television camera control system, characterized in that, The wide-temperature adaptive continuous zoom television camera control system includes a control host and a continuous zoom television camera. The control host is bidirectionally connected to the continuous zoom television camera via a serial communication bus. The control host includes a memory and a processor. The memory stores a computer program that can be executed by the processor. When the processor executes the computer program, it implements the wide-temperature adaptive continuous zoom television camera control method according to any one of claims 1 to 4.

Images