A signal processing method and system for a long-range servo laser seeker

Abstract

This application provides a signal processing method and system for a long-range servo laser seeker, belonging to the field of signal processing technology. This application acquires the target center coordinates, spot image sequence, and synchronized echo pulse waveform sequence of the seeker. It calculates the information entropy of the target spot image and the centroid displacement to obtain the speckle energy entropy and centroid drift, and extracts the pulse width parameter from the echo waveform. These three parameters are used to construct a laser feature vector, which is then input into a pre-trained reflective surface classification model to identify the surface material type of the laser irradiation point in real time. Based on the surface material type, an optimal set of servo control gains is determined from the material gain mapping relationship, and this set is used to dynamically reconstruct the servo control loop. The line-of-sight angle error between the spot centroid and the target center is calculated, and the adjusted control loop is used to calculate this error to generate a precise drive command signal. This application improves the tracking stability and accuracy of the target.

Description

Technical Field

[0001] This application belongs to the field of signal processing technology, and in particular relates to a signal processing method and system for a long-distance servo laser seeker. Background Technology

[0002] Long-range servo laser seekers, as a sophisticated photoelectric detection system, achieve precise tracking of moving targets by processing the laser signals reflected from the target in real time. With increasingly complex modern applications, the stability and accuracy requirements for the signal processing technology of long-range servo laser seekers in complex environments are also rising.

[0003] The signal processing methods of existing laser seekers mainly determine the line-of-sight error by calculating the centroid coordinates of the laser spot, and use an automatic gain control circuit to passively adjust the system gain according to the energy intensity of the received signal, thereby achieving target locking and tracking.

[0004] However, existing technologies have limitations when applied to tracking distant targets with non-uniform surface materials. When the laser illumination point moves rapidly between different material regions of the target, the energy intensity of the echo signal undergoes drastic and unpredictable abrupt changes. Because existing signal processing methods rely solely on passive feedback adjustment of the total signal energy, their control mechanisms suffer from significant response delays, leading to degraded tracking accuracy or even complete loss of lock. Therefore, existing technologies exhibit poor tracking stability and accuracy when applied to tracking distant targets with non-uniform surface materials. Summary of the Invention

[0005] The purpose of this application is to provide a signal processing method, system, electronic device, and storage medium for a long-range servo laser seeker, in order to solve the problems of poor tracking stability and accuracy in the prior art.

[0006] To address the aforementioned technical problems, in a first aspect, this application provides a signal processing method for a long-range servo laser seeker, comprising:

[0007] The target center coordinates of the servo laser seeker, as well as the spot image sequence and echo pulse waveform sequence acquired by the servo laser seeker, are obtained.

[0008] The speckle energy entropy is obtained by calculating the information entropy of the grayscale histogram of the target frame speckle image in the speckle image sequence, and the centroid drift is obtained by calculating the displacement of the centroid coordinates of the target frame speckle image and the adjacent frame speckle images.

[0009] The target echo pulse waveform, which is synchronously acquired with the target frame spot image, is determined from the echo pulse waveform sequence, and the pulse width parameter is extracted from the target echo pulse waveform.

[0010] Laser feature vectors are constructed based on speckle energy entropy, centroid drift, and pulse width parameters. These laser feature vectors are then input into a trained reflective surface classification model to obtain the surface material type of the laser irradiation point. The reflective surface classification model is trained using training samples that include laser feature vectors and corresponding labels.

[0011] Based on the surface material type, the servo control gain set is determined in the preset material gain mapping relationship, and the servo control gain set is used to adjust the servo control loop of the laser seeker. The material gain mapping relationship includes the correspondence between the surface material type and the servo controller gain parameters.

[0012] Based on the target surface center coordinates and the spot centroid coordinates of the target frame spot image, the line-of-sight angle error information is calculated. The adjusted follow-up control loop is then used to process the line-of-sight angle error information to generate a drive command signal for driving the follow-up laser seeker.

[0013] In one feasible implementation, a laser feature vector is constructed based on speckle energy entropy, centroid drift, and pulse width parameters. This laser feature vector is then input into a trained reflective surface classification model to obtain the surface material type of the laser irradiation point, including:

[0014] Laser feature vectors are constructed based on speckle energy entropy, centroid drift, and pulse width parameters.

[0015] A feature vector sequence based on a time sliding window is established. The feature vector sequence includes the laser feature vector and multiple consecutive historical laser feature vectors adjacent to the laser feature vector, which are constructed based on the spot image sequence and the echo pulse waveform sequence before the laser feature vector.

[0016] The mean vector and covariance matrix are calculated based on multiple historical laser eigenvectors in the eigenvector sequence, and the Mahalanobis distance between the laser eigenvector and the historical laser eigenvectors in the eigenvector sequence is calculated based on the mean vector and covariance matrix.

[0017] Determine whether the Mahalanobis distance is less than a preset threshold. If the Mahalanobis distance is less than the preset threshold, input the laser feature vector into the trained reflective surface classification model to obtain the surface material type of the laser irradiation point.

[0018] In one feasible implementation, the method further includes:

[0019] When the Mahalanobis distance is greater than or equal to a preset threshold, the previous historical laser feature vector adjacent to the laser feature vector in the feature vector sequence is input into the trained reflective surface classification model to obtain the surface material type of the laser irradiation point.

[0020] In one feasible implementation, the method further includes:

[0021] The pulse repetition frequency code is decoded by calculating the time interval between adjacent echo pulse waveforms in the echo pulse waveform sequence.

[0022] Based on the surface material type, the servo control gain set is determined in the preset material gain mapping relationship, including:

[0023] The pulse repetition frequency code is matched with the preset code in the servo laser seeker. When the pulse repetition frequency code matches the preset code, the servo control gain set is determined in the preset material gain mapping relationship according to the surface material type.

[0024] In one feasible implementation, the servo laser seeker includes a beam splitter, an imaging detector, and a single-point photodetector.

[0025] The sequence of light spot images is obtained by using a beam splitter to guide a portion of the laser light reflected from the target to an imaging detector for acquisition;

[0026] The echo pulse waveform sequence is obtained by using a beam splitter to guide part of the laser reflected from the tracking target to a single-point photodetector for acquisition and conversion.

[0027] In one feasible implementation, based on the target surface center coordinates and the spot centroid coordinates of the target frame spot image, the line-of-sight angle error information is calculated. The adjusted servo control loop then processes this line-of-sight angle error information to generate a drive command signal for driving the servo laser seeker, including:

[0028] The centroid coordinates of the target frame spot image are vector subtracted from the center coordinates of the target surface to generate a line-of-sight error vector as the line-of-sight angle error information;

[0029] Using the adjusted follow-up control loop, proportional, integral, and derivative operations are performed on the line-of-sight error vector to obtain the proportional control component, integral control component, and derivative control component, respectively. The proportional control component, integral control component, and derivative control component are then linearly superimposed to generate the target control vector.

[0030] The target control vector is converted into a drive command signal for driving the servo laser head.

[0031] In one feasible implementation, before inputting the laser feature vector into a trained reflective surface classification model to obtain the surface material type of the laser irradiation point, the method includes:

[0032] Obtain a training sample set, which includes multiple training samples. Each training sample includes a laser feature vector generated from a sequence of historical spot images and a sequence of historical echo pulse waveforms, and a corresponding label of the real surface material type.

[0033] For each training sample in the training sample set, perform the following steps:

[0034] The laser feature vector in each training sample is input into a pre-defined reflective surface classification model to obtain the predicted surface material type label;

[0035] The loss function value of the reflective surface classification model is determined based on the true surface material type label and the predicted surface material type label of each training sample.

[0036] If the loss function value does not meet the preset training stopping condition, adjust the model parameters of the reflective surface classification model to obtain an updated reflective surface classification model. Then, input the laser feature vector into the reflective surface classification model to obtain the predicted surface material type label. Continue until the loss function value meets the training stopping condition to obtain a trained reflective surface classification model.

[0037] Secondly, this application provides a signal processing system for a long-range servo laser seeker, comprising:

[0038] The acquisition module is used to acquire the target center coordinates of the servo laser seeker, as well as the spot image sequence and echo pulse waveform sequence acquired by the servo laser seeker.

[0039] The calculation module is used to obtain the speckle energy entropy by calculating the information entropy of the gray-level histogram of the target frame speckle image in the speckle image sequence, and to obtain the centroid drift by calculating the displacement of the centroid coordinates of the target frame speckle image and the adjacent frame speckle images.

[0040] The extraction module is used to determine the target echo pulse waveform that is synchronously acquired with the target frame spot image in the echo pulse waveform sequence, and to extract the pulse width parameter from the target echo pulse waveform.

[0041] The classification module is used to construct laser feature vectors based on speckle energy entropy, centroid drift, and pulse width parameters, and input the laser feature vectors into the trained reflective surface classification model to obtain the surface material type of the laser irradiation point. The reflective surface classification model is trained using training samples that include laser feature vectors and corresponding labels.

[0042] The adjustment module is used to determine the servo control gain set in the preset material gain mapping relationship according to the surface material type, and to adjust the servo control loop of the laser seeker using the servo control gain set. The material gain mapping relationship includes the correspondence between the surface material type and the servo controller gain parameters.

[0043] The calculation module is also used to calculate the line-of-sight angle error information based on the target surface center coordinates and the centroid coordinates of the target frame spot image, and to process the line-of-sight angle error information using the adjusted follow-up control loop to generate a drive command signal for driving the follow-up laser seeker.

[0044] Thirdly, this application provides an electronic device, comprising:

[0045] Memory, used to store computer programs;

[0046] A processor is used to execute computer programs to implement the signal processing method of the long-range servo laser seeker as described in the first aspect above.

[0047] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the signal processing steps of the long-range follow-up laser seeker method described in the first aspect above.

[0048] The signal processing method for the long-range servo laser seeker provided in this application acquires a speckle image sequence and an echo pulse waveform sequence, and extracts speckle energy entropy, centroid drift, and pulse width parameters that characterize the spatial and temporal reflection properties of the target surface, thereby constructing a multimodal laser feature vector. Using this feature vector, the surface material type of the current laser irradiation point can be identified in real time through a reflective surface classification model. It can anticipate potential drastic changes in the echo signal caused by material variations and proactively determine the optimal servo control gain set matching the current material characteristics from a preset mapping relationship. Therefore, this application overcomes the control delay problem caused by relying solely on energy feedback in existing technologies. When tracking long-range targets with non-uniform surface materials, it can effectively suppress system oscillations, avoid loss of lock, and improve tracking stability and accuracy.

[0049] Furthermore, this application establishes a time-series window for the laser feature vector and uses Mahalanobis distance to determine the consistency between the current feature vector and historical data distribution in real time. This effectively distinguishes between normal feature changes caused by smooth material transitions and instantaneous abrupt changes caused by abnormal events. When a sudden change is detected, it does not blindly use abnormal data for classification but temporarily uses the reliable classification result from the previous moment, thereby avoiding erroneous decisions and unnecessary controller gain adjustments caused by external instantaneous interference. By introducing a consistency verification mechanism in the time dimension, the reliability of the signal processing method in dynamic interference environments is enhanced, further improving the stability and accuracy of tracking. Attached Figure Description

[0050] To more clearly illustrate the technical solutions of the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 A schematic flowchart illustrating a signal processing method for a long-range servo laser seeker provided in an embodiment of this application;

[0052] Figure 2 A schematic flowchart illustrating another signal processing method for a long-range servo laser seeker provided in an embodiment of this application;

[0053] Figure 3 A flowchart illustrating a method for generating drive instruction signals provided in an embodiment of this application;

[0054] Figure 4 A schematic diagram of the signal processing system for a long-range servo laser seeker provided in this application embodiment;

[0055] Figure 5 This is a schematic diagram of the hardware structure of an electronic device provided in one embodiment of this application. Detailed Implementation

[0056] This application addresses the technical problem in existing laser-guided devices where tracking performance degrades due to the inability to detect changes in target surface material. The core idea of ​​the signal processing method in this application is to move beyond relying solely on single echo energy intensity. Instead, it extracts multiple dimensional features characterizing the physical reflection properties of the target surface, such as speckle energy entropy, centroid drift, and pulse width, from both the two-dimensional spatial domain (spot image sequence) and the one-dimensional time domain (echo pulse waveform sequence). By fusing these multimodal features and utilizing a machine learning model for identification, the target surface material type is predicted. Based on this prediction, the optimal gain parameters are proactively configured for the servo control loop in a feedforward manner, overcoming the inherent delay of traditional passive feedback control and achieving stable and accurate tracking of various targets in complex dynamic scenarios.

[0057] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0058] The core of this application is to provide a signal processing method for a long-range servo laser seeker, and a flowchart of one specific implementation is shown below. Figure 1 As shown, the method includes steps S110 to S160.

[0059] S110: Obtain the target center coordinates of the servo laser seeker, as well as the spot image sequence and echo pulse waveform sequence acquired by the servo laser seeker.

[0060] A servo laser seeker is a precision photoelectric tracking device that achieves stable tracking by receiving the laser energy reflected from the target and driving its optical axis to continuously align with the target. The target center coordinates are a fixed physical point on the target surface of the imaging detector inside the seeker, typically the geometric center of the detector. Precisely calibrated before the system leaves the factory, these coordinates represent the ideal position where the seeker expects the laser spot energy to concentrate, and can be represented as a two-dimensional pixel coordinate. The laser spot image sequence is a series of time-synchronized two-dimensional image frames generated by the imaging detector inside the seeker capturing the laser spot reflected from the target at consecutive time points. Each frame records the spatial distribution and intensity information of the laser energy on the detector target surface at a specific instant. The echo pulse waveform sequence is synchronously acquired by the photoelectric detector inside the seeker, reflecting the one-dimensional time-domain signal reflecting the change in energy intensity of each laser echo pulse over time. Each waveform data in the sequence corresponds precisely to a frame in the laser spot image sequence.

[0061] During the process of a laser seeker locking onto and tracking a distant moving target with a non-uniform surface material, such as a surface vessel or an aircraft, the following steps are taken: First, the pre-calibrated coordinates of the target's center, for example (128, 128), are acquired. These coordinates remain constant throughout the tracking process and serve as a reference for calculating the line-of-sight error. Then, synchronous acquisition commands are sent to the imaging detector and photodetector at a fixed high frequency. Within each acquisition cycle, for example at time t1, when the laser illuminates the diffuse reflection coating area of ​​the target surface, the imaging detector exposes and outputs a digital image frame, such as a 256x256 pixel grayscale image matrix. This image appears as a large spot with a relatively uniform diffuse energy distribution and is stored as the Nth frame in the spot image sequence. Simultaneously, the photodetector and its high-speed sampling circuitry capture the same laser echo pulse and convert the instantaneous change in its intensity into digitized waveform data, such as a one-dimensional array containing 1024 sampling points. This waveform exhibits a moderate amplitude and relatively wide pulse width and is stored as the Nth data frame in the echo pulse waveform sequence.

[0062] This synchronous acquisition process is continuous, generating a dynamic data sequence. For example, at time t2, due to target movement or seeker line-of-sight adjustment, the laser illumination point rapidly moves to the target's transparent window. At this point, the acquisition process repeats; the imaging detector captures and stores the (N+1)th frame image, which appears as a small, extremely bright spot at the center. Simultaneously, the photodetector acquires and stores the (N+1)th echo pulse waveform, which is a sharp pulse with rapidly increasing amplitude and narrowing pulse width. Through this cyclical process, fixed target center coordinates, as well as dynamically changing spot image sequences and echo pulse waveform sequences, are obtained.

[0063] S120: The speckle energy entropy is obtained by calculating the information entropy of the gray-level histogram of the target frame speckle image in the speckle image sequence, and the centroid drift is obtained by calculating the displacement of the centroid coordinates of the target frame speckle image and the adjacent frame speckle images.

[0064] Speckle energy entropy is a scalar parameter used to quantify the uniformity of energy distribution in a speckle image; its magnitude reflects the randomness and complexity of the pixel brightness distribution within the speckle. Centroid drift is a scalar parameter used to characterize the amplitude of minute jitter in the speckle position between consecutive image frames, representing the instantaneous stability of the target's reflectivity.

[0065] For the current target frame of the speckle image sequence, the speckle energy entropy and centroid drift are calculated separately. The speckle energy entropy is calculated by first traversing all pixels of the current image frame and counting the number of pixels at different gray levels to generate a gray-level histogram of the image; then, the frequency of each gray level is substituted into the standard information entropy calculation formula to obtain the final speckle energy entropy value. The centroid drift is calculated by first calculating the brightness-weighted average position of all pixels in the speckle image to determine the precise centroid coordinates representing the energy concentration point of the speckle. This coordinate is a two-dimensional coordinate in pixels; then, the Euclidean distance between this coordinate and the centroid coordinates of the previous frame is calculated to obtain a displacement in pixels, which is the centroid drift.

[0066] For example, when processing a frame of speckle image corresponding to the diffuse reflection coating region of a target, firstly, to calculate the speckle energy entropy, an array of size 256 is created as a grayscale histogram counter. Next, the grayscale value of each pixel in the image matrix of that frame is read one by one, and the corresponding position in the counter array is incremented by one. After traversal, each count value in the counter array is divided by the total number of pixels to obtain the probability of occurrence of each grayscale level. Finally, all probability values ​​are substituted into the information entropy formula for summation to obtain a higher entropy value. Simultaneously, to calculate the centroid drift, two accumulators are initialized to store the weighted sum of the X and Y coordinates, and one accumulator is used to store the total energy. Then, the image matrix is ​​traversed again. For each pixel, its X coordinate is multiplied by its grayscale value and accumulated in the X accumulator, its Y coordinate is multiplied by its grayscale value and accumulated in the Y accumulator, and its grayscale value is accumulated in the total energy accumulator. After traversal, the values ​​of the X and Y accumulators are divided by the total energy value to obtain the centroid coordinates of the current frame. Finally, the centroid coordinates of the previous frame are obtained and the Euclidean distance formula is applied to calculate the distance between the current centroid and the centroid of the previous frame, resulting in a lower centroid drift.

[0067] S130: Determine the target echo pulse waveform that is synchronously acquired with the target frame spot image in the echo pulse waveform sequence, and extract the pulse width parameter from the target echo pulse waveform.

[0068] The target echo pulse waveform is a specific one-dimensional time-domain signal that is precisely synchronized with the currently processed target frame spot image in terms of acquisition time within the echo pulse waveform sequence. The pulse width parameter is a scalar value that quantifies the duration of the target echo pulse waveform on the time axis. It can be calculated by measuring the time interval in which the pulse waveform amplitude exceeds a certain percentage of its peak value, such as the full width at half maximum (FWHM). This parameter reflects the broadening or compression effect of the target surface material on the laser pulse energy in the time dimension.

[0069] By matching the sequence index number of the current target frame spot image, waveform data with the same index number is directly located and extracted from the echo pulse waveform sequence; this is the target echo pulse waveform. Next, this one-dimensional data array is processed to extract the pulse width parameter. First, the maximum amplitude value is determined in the waveform data through peak detection, and then 50% of this peak value is calculated as the half-peak threshold. Subsequently, a search is performed in both forward and backward directions from the peak point to find two time points where the rising and falling edges of the waveform intersect with this threshold. Finally, the time difference between these two time points is calculated to obtain the specific pulse width parameter value, which will be used to subsequently construct the laser feature vector.

[0070] For example, when processing the (N+1)th frame of data, the (N+1)th target echo pulse waveform is retrieved from the echo pulse waveform sequence based on its sequence index. Since the laser irradiation point is located in the glass region, this echo presents as a pulse with a sharply increased amplitude and a sharp shape, and its data is a one-dimensional array containing 1024 sampling points. Scanning this array determines its peak amplitude to be 4.8V. Based on this peak value, the half-peak threshold is calculated to be 2.4V. Subsequently, a search is performed in this array, locating the time point when the waveform rising edge intensity reaches 2.4V at 51.2ns and the time point when the falling edge intensity returns to 2.4V at 51.8ns. Finally, the pulse width parameter is calculated as 0.6ns by subtracting the two. This value is significantly smaller than the 3.2ns pulse width obtained when irradiating the diffuse reflection coating at the previous moment, reflecting the abrupt change in the reflectivity of the target surface.

[0071] S140: A laser feature vector is constructed based on speckle energy entropy, centroid drift, and pulse width parameters. The laser feature vector is then input into a trained reflective surface classification model to obtain the surface material type of the laser irradiation point. The reflective surface classification model is trained using training samples that include the laser feature vector and corresponding labels.

[0072] The laser feature vector is a multi-dimensional array of parameters used to comprehensively and quantitatively describe the characteristics of the current laser echo signal. This vector includes at least three components: speckle energy entropy, centroid drift, and pulse width parameter. The trained reflector classification model is a trained machine learning algorithm model, such as a lightweight neural network or support vector machine. The surface material type is a discrete label output by the reflector classification model, representing the discrimination result of the physical properties of the current laser illumination point, such as highly reflective metal, diffuse reflection coating, composite material, transparent window, and absorbing material.

[0073] First, the speckle energy entropy, centroid drift, and pulse width parameters are combined into a one-dimensional laser feature vector in a predetermined fixed order. Then, this constructed laser feature vector is used as input data and fed into a pre-trained reflective surface classification model for inference calculations. The reflective surface classification model maps the input vector to a predefined material category space and finally outputs the most likely category label, which represents the surface material type of the laser irradiation point at the current moment.

[0074] For example, the N+1th frame of data is processed. For instance, the pre-trained reflective surface classification model can be a trained support vector machine (SVM) classification model, trained to output various predefined surface material types, such as highly reflective metals, diffuse coatings, composite materials, transparent windows, and absorbing materials. First, the speckle energy entropy E(n+1), centroid drift D(n+1), and pulse width parameter W(n+1) calculated in the current frame are combined into a three-dimensional laser feature vector [E(n+1), D(n+1), W(n+1)]. Then, this vector is input into the trained SVM classification model. The model determines that the feature vector belongs to the transparent window category by calculating its position in the feature space relative to the decision boundary learned during training. Therefore, the surface material type result obtained from the N+1th frame of data is a transparent window.

[0075] S150: Based on the surface material type, determine the servo control gain set in the preset material gain mapping relationship, and use the servo control gain set to adjust the servo control loop of the laser seeker. The material gain mapping relationship includes the correspondence between the surface material type and the servo controller gain parameters.

[0076] The servo control gain set refers to a pre-calibrated set of controller parameters designed to handle specific surface material types. It includes three gain values: proportional (P), integral (I), and derivative (D), which determine the response speed, stability, and steady-state accuracy of the servo control loop. The servo control loop is a core closed-loop feedback system within the laser seeker. It receives line-of-sight error signals, processes them through the controller, and generates drive commands to control the rotation of the seeker's mechanical frame to eliminate the errors. The material gain mapping relationship is a pre-established lookup table that corresponds to each identifiable surface material type and an optimal servo control gain set.

[0077] Using the surface material type determined in the previous step as a query index, a search and match is performed within the preset material gain mapping relationship. Upon successful matching, a specific servo control gain set corresponding to that material type is determined. Subsequently, this new set of proportional, integral, and derivative gain parameters is loaded into the controller of the servo control loop in real time, replacing the currently used parameter set. This process achieves dynamic reconstruction of the controller's behavior, enabling its performance characteristics to adapt feedforward to upcoming changes in echo signal characteristics. Ultimately, before processing the line-of-sight error signal of the next frame, the control law of the servo control loop has been adjusted to the state most suitable for the current target surface reflection characteristics.

[0078] For example, once the (N+1)th frame of data is identified as a transparent window, the surface material type label is immediately passed to this step. This label is used as an index to look up the material gain mapping table. An example material gain mapping is shown in Table 1.

[0079] Table 1: Comparison Table of Surface Material Type and Servo Controller Gain Parameters

[0080]

[0081] According to Table 1, the servo control gain set corresponding to the transparent window is {Kp=0.40, Ki=0.10, Kd=0.05}. This set of lower gain values ​​is immediately written into the PID controller register of the servo control loop, replacing the higher gain values ​​previously used to process the diffuse coating signal. Therefore, before calculating and processing the line-of-sight error of the N+1th frame, the response characteristics of the servo control loop have been adjusted to be smoother, preventing system oversaturation, oscillation, or overdrive caused by the strong signal input from the transparent window, thus ensuring the smoothness and stability of the tracking process.

[0082] S160: Based on the target surface center coordinates and the spot centroid coordinates of the target frame spot image, the line-of-sight angle error information is calculated, and the adjusted follow-up control loop is used to process the line-of-sight angle error information to generate a drive command signal for driving the follow-up laser seeker.

[0083] Line-of-sight error information is a two-dimensional vector data representing the deviation between the current line of sight of the seeker and the center line of the target spot. It indicates the offset and direction of the spot centroid relative to the center coordinates of the target surface on the detector surface. Its magnitude and direction correspond to the angle and direction that the seeker's line of sight needs to be adjusted. The drive command signal is the final output generated after processing by the servo control loop. It is a set of electrical signals, which can be voltage or pulse width modulation signals, used to directly drive the servo motor or torque converter of the seeker, making it rotate at a specified angular velocity or angle, thereby eliminating the line-of-sight error.

[0084] By subtracting the centroid coordinates of the current target frame's spot image from the pre-calibrated target surface center coordinates, a two-dimensional error vector is obtained, which represents the line-of-sight angle error information. This error vector is then input into a servo control loop whose gain has been adjusted according to the surface material type. Inside this loop, the PID controller uses new proportional, integral, and derivative gain parameters to perform real-time calculations on the input error vector, generating a target control vector. Finally, this target control vector is converted into a physical electrical signal—the drive command signal—that can directly drive the seeker frame to rotate, thus completing one closed-loop control cycle.

[0085] Figure 2 The diagram shows a schematic flow chart of another signal processing method for a long-range servo laser seeker provided in one embodiment of this application.

[0086] In one feasible implementation, such as Figure 2 As shown, step S140: Construct a laser feature vector based on speckle energy entropy, centroid drift, and pulse width parameters, and input the laser feature vector into the trained reflective surface classification model to obtain the surface material type of the laser irradiation point, including:

[0087] S141: Construct laser feature vectors based on speckle energy entropy, centroid drift, and pulse width parameters.

[0088] The calculated speckle energy entropy, centroid drift, and pulse width parameters are arranged in a pre-defined fixed order and combined into a laser feature vector.

[0089] S142: Establish a feature vector sequence based on a time sliding window. The feature vector sequence includes the laser feature vector and multiple consecutive historical laser feature vectors adjacent to the laser feature vector, constructed based on the spot image sequence and echo pulse waveform sequence before the laser feature vector.

[0090] A time-sliding window is a data processing mechanism used to maintain a fixed-length, chronologically ordered data sequence. As new data arrives, the oldest data is removed, causing the window to slide forward along the timeline. The feature vector sequence is the collection of all laser feature vectors stored within the time-sliding window at any given moment. Historical laser feature vectors refer to all previously acquired feature vectors in the feature vector sequence, excluding the most recently constructed laser feature vector.

[0091] A first-in-first-out queue or circular buffer is established as a time-sliding window, with its length N preset. Whenever a new laser feature vector is generated, it is added to the end of the window. If the number of vectors in the window exceeds N, the oldest historical laser feature vector at the front of the window is discarded. In this way, the window always stores the laser feature vectors corresponding to the current frame and the N-1 consecutive historical frames preceding it, collectively forming the latest feature vector sequence.

[0092] S143: Calculate the mean vector and covariance matrix based on multiple historical laser eigenvectors in the eigenvector sequence, and calculate the Mahalanobis distance between the laser eigenvector and the historical laser eigenvectors in the eigenvector sequence based on the mean vector and covariance matrix.

[0093] The mean vector is obtained by averaging the corresponding dimensional components of all vectors in the eigenvector sequence. In the covariance matrix, the diagonal elements represent the variance of each dimension, and the off-diagonal elements represent the covariance between different dimensions. Mahalanobis distance is a distance metric that considers the correlation of data distributions; it represents the statistical distance between a data point and a data distribution center.

[0094] First, all historical laser eigenvectors are extracted from the eigenvector sequence. Then, their mean vector and covariance matrix are calculated using these historical vectors. These two statistics together describe the stable distribution of the recent target reflection characteristics. Next, the latest laser eigenvector, along with the calculated mean vector and covariance matrix, are substituted into the standard formula for calculating Mahalanobis distance to obtain a non-negative scalar value, namely the Mahalanobis distance. The formula for calculating Mahalanobis distance is shown in formula (1).

[0095] (1)

[0096] in, Represents Mahalanobis distance, This represents the current laser feature vector, i.e., the data point whose distance needs to be measured. The mean vector representing the set of historical laser feature vectors. The covariance matrix represents the set of historical laser eigenvectors. Represents the covariance matrix The inverse matrix.

[0097] S1441: Determine whether the Mahalanobis distance is less than a preset threshold. If the Mahalanobis distance is less than the preset threshold, input the laser feature vector into the trained reflective surface classification model to obtain the surface material type of the laser irradiation point.

[0098] The calculated Mahalanobis distance value is compared with a preset threshold. If the Mahalanobis distance is less than the threshold, it indicates that the current laser feature vector is statistically consistent with recent historical data and does not belong to a sudden change or abnormal signal. In this case, the latest laser feature vector is used as a valid input and fed into a trained reflective surface classification model for classification. The output of the model after inference is the surface material type of the current laser irradiation point.

[0099] For example, assume the length of the time sliding window is set to 10. During the continuous and stable tracking of the diffuse coating region of the target ship by the laser seeker, 9 historical laser feature vectors are stored within the sliding window. When processing the Nth frame of data, the calculated speckle energy entropy En, centroid drift Dn, and pulse width Wn are first combined into a new laser feature vector Vn=[En, Dn, Wn]. Vn is added to the window, which now contains 10 feature vectors from the (N-9)th frame to the Nth frame, forming the latest feature vector sequence. The 9 historical vectors from the (N-9)th frame to the (N-1)th frame are extracted from the window, and their mean vector μ and covariance matrix S are calculated. Since the same material is being tracked during this period, these 9 vectors are compactly distributed. Subsequently, the Mahalanobis distance between the current vector Vn and this distribution defined by the historical data is calculated. Because the Nth frame still illuminates the diffuse coating, the value of Vn will be very close to the historical mean μ, so the calculated Mahalanobis distance value will be very small, for example, 0.8. Finally, the Mahalanobis distance of 0.8 is compared with a preset threshold, such as 3.0. A Mahalanobis distance less than the preset threshold indicates signal stability. Therefore, the laser feature vector Vn is input into the trained support vector machine classification model. Based on the feature values ​​of Vn, the model determines whether it belongs to the diffuse reflection coating category and outputs the corresponding material type label.

[0100] In one feasible implementation, such as Figure 2 As shown, step S140 of the method further includes:

[0101] S1442: When the Mahalanobis distance is greater than or equal to a preset threshold, the previous historical laser feature vector adjacent to the laser feature vector in the feature vector sequence is input into the trained reflective surface classification model to obtain the surface material type of the laser irradiation point.

[0102] When the calculated Mahalanobis distance is not less than a preset threshold, the latest laser feature vector is ignored. Instead, the penultimate vector, the historical laser feature vector adjacent to the current vector, is located and extracted from the feature vector sequence maintained by the time sliding window. This extracted historical vector is then used as input to a trained reflective surface classification model for inference. The classification label output by the model is used as the surface material type of the laser-irradiated point at the current moment. The last reliable classification result temporarily replaces the current unreliable one to enhance the system's stability during sudden state changes.

[0103] This embodiment establishes a time-series window for the laser feature vector and uses Mahalanobis distance to determine the consistency between the current feature vector and historical data distribution in real time. This effectively distinguishes between normal feature changes caused by smooth material transitions and instantaneous abrupt changes caused by abnormal events. When a sudden change is detected, it does not blindly use abnormal data for classification but temporarily uses the reliable classification result from the previous moment, thereby avoiding erroneous decisions and unnecessary controller gain adjustments caused by external instantaneous interference. By introducing a consistency verification mechanism in the time dimension, the reliability of the signal processing method in dynamic interference environments is enhanced, further improving the stability and accuracy of tracking.

[0104] In one feasible implementation, the method further includes:

[0105] The pulse repetition frequency code is decoded by calculating the time interval between adjacent echo pulse waveforms in the echo pulse waveform sequence.

[0106] Pulse repetition frequency coding is a technique that transmits specific information or authentication codes by precisely controlling the time interval sequence of laser pulse emission. It is a specific pattern composed of multiple consecutive different time intervals. The time interval refers to the length of time elapsed between two adjacent echo pulse waveforms received by the sensor.

[0107] Decoding the pulse repetition frequency code involves restoring the received time interval sequence into a predefined encoding format. Using a high-precision clock inside the seeker, a precise arrival timestamp is recorded for each detected echo pulse waveform. By continuously calculating the difference between the timestamp of the current pulse and the timestamp of the previous pulse, a time interval sequence composed of actual measurements can be obtained. This measurement sequence is compared and quantized against preset encoding rules, such as mapping a specific range of time intervals to the encoded symbols 0 or 1, thereby converting an analog time sequence into a digital encoded sequence. This sequence is the decoded pulse repetition frequency code.

[0108] Based on the surface material type, the servo control gain set is determined in the preset material gain mapping relationship, including:

[0109] The pulse repetition frequency code is matched with the preset code in the servo laser seeker. When the pulse repetition frequency code matches the preset code, the servo control gain set is determined in the preset material gain mapping relationship according to the surface material type.

[0110] The preset code is a specific pulse repetition frequency encoding sequence that is pre-constructed and serves as a password or key for the seeker to identify legitimate laser irradiation sources.

[0111] The matching process involves comparing the real-time decoded pulse repetition frequency code with this preset code bit by bit or in its entirety to verify their consistency. First, the pulse repetition frequency code decoded in the previous step is compared with the preset code stored internally by the seeker. Correlation operations or pattern matching algorithms can be used to determine if they match within a certain tolerance range. If the comparison results in a successful match, authentication is passed. Based on the identified surface material type, the corresponding servo control gain set is located and determined in the material gain mapping relationship, and used to adjust the servo control loop. Conversely, if the codes do not match, it indicates that the currently received laser signal is not from an authorized irradiation source. This will prevent subsequent material-based gain adjustments to prevent interference from the enemy or accidental irradiation from friendly forces, thus ensuring that the seeker only responds to correct command signals.

[0112] For example, suppose the seeker is configured to respond only to laser illumination using a specific pulse repetition frequency code, such as Code-1234, which is stored as a preset code within the seeker. In a tracking mission, a friendly laser illuminator illuminates the target ship using Code-1234. During tracking, the seeker measures the time interval between consecutive echo pulses, decodes the current pulse repetition frequency code as Code-1234, and successfully matches it with the internal preset code, thus verifying authentication. If the seeker's line of sight moves from the ship's diffuse coating to a transparent window, it identifies the material type as transparent. Since the code has matched, the control flow can continue. Based on the transparent window material type, the corresponding gain parameter set is retrieved from the material gain mapping table and immediately updated with it to update the servo control loop. If, in another scenario, a laser source using an unauthorized code, Code-5678, illuminates the target, the seeker's decoding fails to match, and even if the material is identified, subsequent gain adjustment steps are prevented, ensuring the seeker only responds to authorized signals and guaranteeing security.

[0113] In one feasible implementation, the servo laser seeker includes a beam splitter, an imaging detector, and a single-point photodetector. A sequence of laser spot images is obtained by guiding a portion of the laser reflected from the tracking target to the imaging detector via the beam splitter for acquisition. A sequence of echo pulse waveforms is obtained by guiding a portion of the laser reflected from the tracking target to the single-point photodetector via the beam splitter for acquisition and conversion.

[0114] A beam splitter is an optical element that splits a beam of incident light into two or more beams according to a specific energy ratio, allowing them to propagate along different optical paths. An imaging detector is a planar photoelectric sensor, such as a charge-coupled device (CCD) or complementary metal-oxide-semiconductor (CMOS) sensor. It consists of multiple independent pixel units arranged to capture the spatial distribution information of light energy projected onto its surface and convert it into a two-dimensional digital image. A single-point photodetector is a photoelectric sensor without spatial resolution, such as an avalanche photodiode (APD) or a PIN photodiode. It has only one photosensitive area used to measure the change in total light power incident on it over time, and is characterized by extremely fast response speed.

[0115] The laser echo signal reflected from a distant target is first focused by the forward optical system of the seeker. In the optical path, this focused laser beam encounters a beam splitter. The beam splitter, according to its preset splitting ratio, transmits or reflects a portion of the laser energy onto a dedicated optical path, the endpoint of which is the imaging detector. When this portion of the laser energy illuminates the photosensitive surface of the imaging detector, the detector, triggered by a synchronization signal, performs an exposure, converting the two-dimensional spatial intensity distribution of the light spot into the charge of each pixel. This data is then processed by a readout circuit and an analog-to-digital converter to generate a digitized grayscale image. This process is repeated continuously at a fixed frequency, thus acquiring a series of temporally continuous light spot images, forming a light spot image sequence.

[0116] Figure 3 A flowchart illustrating a method for generating drive instruction signals according to an embodiment of this application is shown. Figure 3 As shown, the scheme includes steps S310 to S330.

[0117] In one feasible implementation, based on the target surface center coordinates and the spot centroid coordinates of the target frame spot image, the line-of-sight angle error information is calculated. The adjusted servo control loop then processes this line-of-sight angle error information to generate a drive command signal for driving the servo laser seeker, including:

[0118] S310: Subtract the centroid coordinates of the target frame spot image from the target surface center coordinates to generate a line-of-sight error vector as the line-of-sight angle error information.

[0119] The gaze error vector is a two-dimensional vector. Its two components correspond to the pixel offset in the horizontal and vertical directions, respectively. Its direction points to the direction that needs to be corrected, and its magnitude is proportional to the angle that needs to be corrected.

[0120] First, obtain the centroid coordinates of the current target frame's spot image. and the pre-calibrated target center coordinates Then, a line-of-sight error vector is generated by performing vector subtraction. ,Right now For example, when processing the (N+1)th frame of data illuminating the transparent window, if the centroid coordinates of the light spot are (132, 126) and the target center coordinates are (128, 128), then the line-of-sight error vector generated in this step is... The pixel vector will serve as the input for the next step of the servo control loop.

[0121] S320: Using the adjusted follow-up control loop, proportional, integral and derivative operations are performed on the line-of-sight error vector to obtain the proportional control component, integral control component and derivative control component respectively. The proportional control component, integral control component and derivative control component are then linearly superimposed to generate the target control vector.

[0122] The proportional control component is a direct response to the current error, and its magnitude is proportional to the error. The integral control component is used to eliminate the steady-state error of the system; it accumulates past errors. The derivative control component is used to predict the trend of error changes. The target control vector is a control command calculated by the PID controller based on the current and historical errors; it indicates the magnitude and direction of the response that the actuator needs to make.

[0123] The generated line-of-sight error vector As input to the PID controller, the controller utilizes the dynamically adjusted gain parameters. , , The error is calculated to generate the target control vector. Its ideal continuous-time control law is shown in formula (2).

[0124] (2)

[0125] in, Represents the target control vector. , , These represent proportional, integral, and differential gains, respectively. This represents the line-of-sight error vector that varies over time.

[0126] For example, the controller is currently using a low gain set {Kp=0.40, Ki=0.10, Kd=0.05} set for the transparent window. The gaze error vector... After input, due to the low gain, the calculated target control vector The amplitude will be moderately suppressed, thereby generating a smooth control command and avoiding overdrive due to errors under strong signals.

[0127] S330: Converts the target control vector into a drive command signal for driving the servo laser head.

[0128] Drive command signals are electrical signals that can be directly recognized and executed by the physical actuators of the guide head, such as servo motors.

[0129] The conversion process translates the calculated digitized target control vector into specific voltage or pulse signals with physical meaning. The generated target control vector... It is a two-dimensional vector, whose components and These correspond to the control quantities for the azimuth and pitch axes, respectively. At a specific moment in the current frame, these two control components... and The instantaneous values ​​are fed into two independent digital-to-analog converters (DACs) or pulse-width modulation (PWM) signal generators. These hardware circuits convert the digital control quantities into corresponding analog voltages or pulse signals with specific duty cycles. For example, the components of the target control vector generated in the previous step can be converted into two voltage signals. These two ultimately generated electrical signals, which are the drive command signals, are directly applied to the seeker's drive circuitry, driving the azimuth and pitch motors to rotate smoothly, accurately and without overshoot to eliminate line-of-sight errors.

[0130] In one feasible implementation, before inputting the laser feature vector into a trained reflective surface classification model to obtain the surface material type of the laser irradiation point, the method includes:

[0131] Obtain a training sample set, which includes multiple training samples. Each training sample includes a laser feature vector generated from a sequence of historical spot images and a sequence of historical echo pulse waveforms, and a corresponding label of the real surface material type.

[0132] First, using the same equipment as the seeker, standard samples of various known materials are laser-irradiated, and a large number of spot images and echo waveform data are acquired. Then, the feature extraction methods described in S120 and S130 are applied to these raw data to generate a large number of laser feature vectors. Finally, all vectors generated from the sample data of specific materials are batch-assigned with their corresponding real material labels, thereby constructing a complete labeled training sample set.

[0133] For each training sample in the training sample set, perform the following steps:

[0134] The laser feature vector from each training sample is input into a pre-defined reflective surface classification model to obtain a predicted surface material type label. Based on the true and predicted surface material type labels of each training sample, the loss function value of the reflective surface classification model is determined. If the loss function value does not meet the pre-defined training stopping condition, the model parameters of the reflective surface classification model are adjusted to obtain an updated reflective surface classification model. The process continues until the loss function value meets the training stopping condition, resulting in a trained reflective surface classification model.

[0135] First, initialize a reflective surface classification model, such as a lightweight neural network or support vector machine (SVM). Then, begin an iterative training loop: take a sample from the training set, input its laser feature vector into the model for a forward propagation calculation, and obtain a predicted surface material type; next, compare this prediction result with the sample's true label, and calculate the current prediction loss value using loss functions such as cross-entropy; then, check if the training has reached the stopping condition, such as reaching a preset number of training epochs or the loss value falling below a certain threshold. If not, use the loss value to fine-tune the model's internal parameters using backpropagation and gradient descent optimizers to minimize the loss in the next prediction. After parameter updates, process the next training sample and repeat the above process. This iterative process traverses the entire training set multiple times until the training stopping condition is met. After training, a series of parameters within the model are finally determined, and the model at this point is the trained reflective surface classification model.

[0136] The signal processing method for the long-range servo laser seeker provided in this application acquires a speckle image sequence and an echo pulse waveform sequence, and extracts speckle energy entropy, centroid drift, and pulse width parameters that characterize the spatial and temporal reflection properties of the target surface, thereby constructing a multimodal laser feature vector. Using this feature vector, the surface material type of the current laser irradiation point can be identified in real time through a reflective surface classification model. It can anticipate potential drastic changes in the echo signal caused by material variations and proactively determine the optimal servo control gain set matching the current material characteristics from a preset mapping relationship. Therefore, this application overcomes the control delay problem caused by relying solely on energy feedback in existing technologies. When tracking long-range targets with non-uniform surface materials, it can effectively suppress system oscillations, avoid loss of lock, and improve tracking stability and accuracy.

[0137] Figure 4 This is a schematic diagram illustrating a specific embodiment of a signal processing system for a long-range servo laser seeker provided in this application. (Refer to...) Figure 4 The system may include:

[0138] The acquisition module 410 is used to acquire the target surface center coordinates of the servo laser seeker, as well as the spot image sequence and echo pulse waveform sequence acquired by the servo laser seeker.

[0139] The calculation module 420 is used to obtain the speckle energy entropy by calculating the information entropy of the gray-level histogram of the target frame speckle image in the speckle image sequence, and to obtain the centroid drift by calculating the displacement of the centroid coordinates of the target frame speckle image and the adjacent frame speckle images.

[0140] Extraction module 430 is used to determine the target echo pulse waveform that is synchronously acquired with the target frame spot image in the echo pulse waveform sequence, and extract the pulse width parameter from the target echo pulse waveform;

[0141] The classification module 440 is used to construct a laser feature vector based on speckle energy entropy, centroid drift and pulse width parameters, and input the laser feature vector into the trained reflective surface classification model to obtain the surface material type of the laser irradiation point. The reflective surface classification model is trained using training samples including laser feature vector and corresponding labels.

[0142] The adjustment module 450 is used to determine the servo control gain set in the preset material gain mapping relationship according to the surface material type, and to adjust the servo control loop of the laser seeker using the servo control gain set. The material gain mapping relationship includes the correspondence between the surface material type and the servo controller gain parameters.

[0143] The calculation module 420 is also used to calculate the line-of-sight angle error information based on the target surface center coordinates and the spot centroid coordinates of the target frame spot image, and to process the line-of-sight angle error information using the adjusted follow-up control loop to generate a drive command signal for driving the follow-up laser seeker.

[0144] In one embodiment, the classification module 440 is specifically used to construct a laser feature vector based on speckle energy entropy, centroid drift, and pulse width parameters; establish a feature vector sequence based on a time sliding window, the feature vector sequence including the laser feature vector and multiple consecutive historical laser feature vectors adjacent to the laser feature vector constructed before the laser feature vector based on the speckle image sequence and echo pulse waveform sequence; calculate the mean vector and covariance matrix based on the multiple historical laser feature vectors in the feature vector sequence, and calculate the Mahalanobis distance between the laser feature vector and the historical laser feature vectors in the feature vector sequence according to the mean vector and covariance matrix; determine whether the Mahalanobis distance is less than a preset threshold, and if the Mahalanobis distance is less than the preset threshold, input the laser feature vector into the trained reflective surface classification model to obtain the surface material type of the laser irradiation point.

[0145] In one embodiment, the classification module 440 is further configured to input the previous historical laser feature vector adjacent to the laser feature vector in the feature vector sequence into the trained reflective surface classification model when the Mahalanobis distance is greater than or equal to a preset threshold, so as to obtain the surface material type of the laser irradiation point.

[0146] In one embodiment, the adjustment module 450 is further configured to decode the pulse repetition frequency code by calculating the time interval between adjacent echo pulse waveforms in the echo pulse waveform sequence; match the pulse repetition frequency code with a preset code in the servo laser seeker; and, if the pulse repetition frequency code matches the preset code, determine the servo control gain set in a preset material gain mapping relationship according to the surface material type.

[0147] In one embodiment, the servo laser seeker includes a beam splitter, an imaging detector, and a single-point photodetector; the beam spot image sequence is obtained by guiding a portion of the laser reflected from the tracking target to the imaging detector for acquisition through the beam splitter; the echo pulse waveform sequence is obtained by guiding a portion of the laser reflected from the tracking target to the single-point photodetector for acquisition and conversion through the beam splitter.

[0148] In one embodiment, the calculation module 420 is further configured to perform vector subtraction between the centroid coordinates of the target frame spot image and the center coordinates of the target surface to generate a line-of-sight error vector as line-of-sight angle error information; use the adjusted follower control loop to perform proportional, integral and differential operations on the line-of-sight error vector to obtain proportional control components, integral control components and differential control components respectively, and linearly superimpose the proportional control components, integral control components and differential control components to generate a target control vector; convert the target control vector into a drive command signal for driving the follower laser head.

[0149] In one embodiment, the classification module 440 is further configured to, before inputting the laser feature vector into the trained reflective surface classification model to obtain the surface material type of the laser irradiation point, acquire a training sample set, the training sample set including multiple training samples, each training sample including a laser feature vector generated from a historical spot image sequence and a historical echo pulse waveform sequence and a corresponding real surface material type label; for each training sample in the training sample set, perform the following steps respectively: input the laser feature vector in each training sample into a preset reflective surface classification model to obtain a predicted surface material type label; determine the loss function value of the reflective surface classification model based on the real surface material type label and the predicted surface material type label of each training sample; if the loss function value does not meet the preset training stopping condition, adjust the model parameters of the reflective surface classification model to obtain an updated reflective surface classification model, and return to input the laser feature vector into the reflective surface classification model to obtain a predicted surface material type label, until the loss function value meets the training stopping condition, and obtain a trained reflective surface classification model.

[0150] The signal processing system of the long-range servo laser seeker in this application embodiment is used to implement the aforementioned signal processing method of the long-range servo laser seeker. Therefore, the specific implementation of the signal processing system of the long-range servo laser seeker can be found in the embodiment section of the signal processing method of the long-range servo laser seeker above. The specific implementation can be referred to the description of the corresponding embodiments, and will not be repeated here.

[0151] Figure 5 A schematic diagram of the hardware structure of an electronic device provided in one embodiment of this application is shown.

[0152] The electronic device may include a processor 510 and a memory 520 storing computer program instructions.

[0153] Specifically, the processor 510 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0154] Memory 520 may include mass storage for data or instructions. For example, and not limitingly, memory 520 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 520 may include removable or non-removable (or fixed) media. Where appropriate, memory 520 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 520 is non-volatile solid-state memory.

[0155] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to the first aspect of this disclosure.

[0156] The processor 510 reads and executes computer program instructions stored in the memory 520 to implement any of the signal processing methods for the long-range servo laser seeker in the above embodiments.

[0157] In one example, the electronic device may also include a communication interface 530 and a bus 540. Wherein, such as Figure 5 As shown, the processor 510, memory 520, and communication interface 530 are connected through bus 540 and complete communication with each other.

[0158] The communication interface 530 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0159] Bus 540 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 540 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0160] The electronic device can execute the signal processing method of the long-range servo laser seeker in the embodiments of this application, thereby realizing the signal processing method of the long-range servo laser seeker described in conjunction with the accompanying drawings.

[0161] Furthermore, in conjunction with the signal processing method of the long-range servo laser seeker in the above embodiments, this application embodiment can provide a computer-readable storage medium for implementation. This computer-readable storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the signal processing methods of the long-range servo laser seeker in the above embodiments.

[0162] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0163] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0164] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0165] The aspects of this application have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by dedicated hardware performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0166] The signal processing method, system, electronic device, and storage medium for a long-range servo laser seeker provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this application. It should be noted that those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A signal processing method for a long-range homing laser seeker, characterized by, The method comprises: acquiring a target surface center coordinate of a tracking laser seeker, and a sequence of spot images and a sequence of echo pulse waveforms collected by the tracking laser seeker; obtaining a speckle energy entropy by calculating an information entropy of a gray histogram of a target frame spot image in the sequence of spot images, and obtaining a centroid drift amount by calculating a displacement amount of a spot centroid coordinate of the target frame spot image and a neighboring frame spot image; determining a target echo pulse waveform synchronously collected with the target frame spot image in the sequence of echo pulse waveforms, and extracting a pulse width parameter from the target echo pulse waveform; constructing a laser feature vector based on the speckle energy entropy, the centroid drift amount and the pulse width parameter, and inputting the laser feature vector into a trained reflector classification model to obtain a surface material type of a laser irradiation point, the reflector classification model being trained by using training samples comprising laser feature vectors and corresponding labels; determining a tracking control gain set in a preset material gain mapping relationship according to the surface material type, and adjusting a tracking control loop of the laser seeker by using the tracking control gain set, the material gain mapping relationship comprising a corresponding relationship between surface material types and tracking controller gain parameters; calculating a line-of-sight angle error information based on the target surface center coordinate and the spot centroid coordinate of the target frame spot image, and performing operation processing on the line-of-sight angle error information by using the adjusted tracking control loop to generate a driving instruction signal for driving the tracking laser seeker.

2. The method of claim 1, wherein, The method further comprises: constructing the laser feature vector based on the speckle energy entropy, the centroid drift amount and the pulse width parameter; establishing a feature vector sequence based on a time sliding window, the feature vector sequence comprising the laser feature vector and a plurality of continuous historical laser feature vectors adjacent to the laser feature vector and constructed before the laser feature vector based on the sequence of spot images and the sequence of echo pulse waveforms; calculating a mean vector and a covariance matrix based on the plurality of historical laser feature vectors of the feature vector sequence, and calculating a Mahalanobis distance between the laser feature vector and the historical laser feature vectors in the feature vector sequence according to the mean vector and the covariance matrix; determining whether the Mahalanobis distance is less than a preset threshold, and inputting the laser feature vector into the trained reflector classification model to obtain the surface material type of the laser irradiation point in a case where the Mahalanobis distance is less than the preset threshold.

3. The method of claim 2, wherein, The method further comprises: in a case where the Mahalanobis distance is greater than or equal to the preset threshold, inputting a previous historical laser feature vector adjacent to the laser feature vector in the feature vector sequence into the trained reflector classification model to obtain the surface material type of the laser irradiation point.

4. The method of claim 1, wherein, The method further comprises: decoding a pulse repetition frequency code by calculating a time interval between adjacent echo pulse waveforms in the sequence of echo pulse waveforms; determining a servo control gain set in a preset material gain mapping relationship according to the surface material type, comprises: matching the pulse repetition frequency code with a preset code in the servo laser seeker, and determining the servo control gain set in the preset material gain mapping relationship according to the surface material type in the case that the pulse repetition frequency code matches the preset code.

5. The method of claim 1, wherein, The servo laser seeker comprises a beam splitter, an imaging detector and a single-point photoelectric detector; The sequence of spot images is obtained by guiding part of the laser reflected by the tracking target to the imaging detector through the beam splitter for collection; The sequence of echo pulse waveforms is obtained by guiding part of the laser reflected by the tracking target to the single-point photoelectric detector through the beam splitter for collection and conversion.

6. The method of claim 1, wherein, The line-of-sight angle error information is calculated based on the target surface center coordinates and the spot centroid coordinates of the target frame spot image, and the line-of-sight angle error information is processed by the adjusted servo control loop to generate a drive instruction signal for driving the servo laser seeker, comprising: Vector subtraction is performed between the spot centroid coordinates of the target frame spot image and the target surface center coordinates to generate a line-of-sight error vector as the line-of-sight angle error information; The line-of-sight error vector is processed by the adjusted servo control loop to obtain proportional control components, integral control components and differential control components, respectively, and the proportional control components, the integral control components and the differential control components are linearly superimposed to generate a target control vector; The target control vector is converted into the drive instruction signal for driving the servo laser seeker.

7. The method of claim 1, wherein, Before the laser feature vector is input into the trained reflector classification model to obtain the surface material type of the laser irradiation point, the method comprises: obtaining a training sample set, the training sample set comprising a plurality of training samples, each training sample comprising a laser feature vector generated from a historical sequence of spot images and a historical sequence of echo pulse waveforms and a corresponding real surface material type label; for each training sample in the training sample set, the following steps are performed respectively: inputting the laser feature vector in each training sample into a preset reflector classification model to obtain a predicted surface material type label; determining a loss function value of the reflector classification model according to the real surface material type label and the predicted surface material type label of each training sample; in the case that the loss function value does not satisfy a preset training stop condition, adjusting the model parameters of the reflector classification model to obtain an updated reflector classification model, and returning to input the laser feature vector into the reflector classification model to obtain the predicted surface material type label until the loss function value satisfies the training stop condition to obtain the trained reflector classification model.

8. A signal processing system for a long-range homing laser seeker, characterized by The system comprises: An acquisition module, configured to acquire a target surface center coordinate of a tracking laser seeker, and a sequence of spot images and a sequence of echo pulse waveforms collected by the tracking laser seeker; A calculation module, configured to obtain a speckle energy entropy by calculating an information entropy of a gray histogram of a target frame spot image in the sequence of spot images, and obtain a centroid drift amount by calculating a displacement amount of a spot centroid coordinate of the target frame spot image and a neighboring frame spot image; An extraction module, configured to determine a target echo pulse waveform collected synchronously with the target frame spot image in the sequence of echo pulse waveforms, and extract a pulse width parameter from the target echo pulse waveform; A classification module, configured to construct a laser feature vector based on the speckle energy entropy, the centroid drift amount, and the pulse width parameter, and input the laser feature vector into a trained reflective surface classification model to obtain a surface material type of a laser irradiation point, the reflective surface classification model being trained by using training samples comprising laser feature vectors and corresponding labels; An adjustment module, configured to determine a tracking control gain set in a preset material gain mapping relationship according to the surface material type, and adjust a tracking control loop of the laser seeker by using the tracking control gain set, the material gain mapping relationship comprising a corresponding relationship between surface material types and tracking controller gain parameters; The calculation module is further configured to calculate a line-of-sight angle error information based on the target surface center coordinate and a spot centroid coordinate of the target frame spot image, and perform operation processing on the line-of-sight angle error information by using the adjusted tracking control loop to generate a driving instruction signal for driving the tracking laser seeker.

9. An electronic device, comprising: The electronic device comprises: A memory, configured to store a computer program; A processor, configured to implement the steps of the signal processing method of the long-distance tracking laser seeker according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program can implement the signal processing method of the long-distance tracking laser seeker according to any one of claims 1 to 7 when executed by a processor.

Images