A target motion trajectory output method, system, device and medium

Abstract

The application relates to the technical field of trajectory analysis, and provides a target motion trajectory output method, system, device and medium. The output method comprises the following steps: initializing a neural synapse array; collecting an optical image of a target scene, and projecting the optical image to the initialized neural synapse array to obtain an initial pixel state set corresponding to the target scene; processing the initial pixel state set to determine a target region and a background region; irradiating the target region and the background region respectively to obtain an updated pixel state set; generating a current attention graph according to the updated pixel state set; and outputting a target motion trajectory according to the current attention graph. The output of the target motion trajectory is realized through a full-light control neural synapse array, so that the system power consumption and processing delay can be significantly reduced while the high real-time performance is maintained.

Description

Technical Field

[0001] This application relates to the field of trajectory analysis technology, and in particular to a method, system, device and medium for outputting target motion trajectory. Background Technology

[0002] Most existing target tracking systems follow a "frame-based imaging plus electronic computing" technical approach: the scene is first output as continuous image frames by imaging devices such as CMOS or CCD. These image frames undergo denoising, de-pixelation, white balance, color and distortion correction, etc., through the image signal processing link inside the camera. The entire frame image data is then transmitted to a back-end general-purpose processor or dedicated accelerator to run algorithms such as correlation filtering and deep learning for target localization and trajectory updates. This "image first, then transmit, then compute" pipeline architecture has been widely adopted in the field of computer vision. To improve robustness under occlusion, deformation, and background interference, existing solutions often continuously increase network size and computational complexity. This is feasible for desktop computing power, but it is prone to phenomena such as decreased inference frame rate, latency jitter, and thermal limitation frequency reduction on low-power devices, making it difficult to guarantee the real-time performance and predictability of tracking output. Summary of the Invention

[0003] This application aims to improve at least one technical problem in the background art.

[0004] This application provides a method for outputting a target motion trajectory, which includes initializing a neural synapse array; Acquire optical images of the target scene and project them onto an initialized neural synapse array to obtain an initial set of pixel states corresponding to the target scene; The initial pixel state set is processed to determine the target region and the background region; The target region and the background region are illuminated separately to obtain an updated pixel state set; Generate the current attention map based on the updated pixel state set; Output the target's motion trajectory based on the current attention map.

[0005] According to some technical solutions of this application, the step of acquiring optical images of the target scene and projecting the optical images onto an initialized neural synapse array to obtain an initial pixel state set corresponding to the target scene specifically includes: At least two frames of optical images of the target scene are continuously acquired to obtain a set of adjacent frame images; Differential processing is performed on adjacent frame image sets to obtain differential images; Extract the grayscale difference of each pixel in the difference image as the intensity change, and compare the intensity change of each pixel with a preset change threshold; If the intensity change is greater than the change threshold, it is identified as a possible target area; The potential target area is illuminated, and the illuminated potential target area is integrated with the remaining unilluminated areas to obtain an initial pixel state set containing all pixel state values.

[0006] According to some technical solutions of this application, the process of processing the initial pixel state set to determine the target region and the background region specifically includes: Extract the state values ​​of each pixel from the initial pixel state set; The state value of each pixel in the initial pixel state set is compared with a preset first threshold. If the state value is greater than the first threshold, the pixel is determined to be the target region. The state value of each pixel in the initial pixel state set is compared with a preset second threshold. If the state value is less than the second threshold, the pixel is determined to be a background area. The judgment results are integrated to obtain the target area and the background area.

[0007] According to some technical solutions of this application, the step of illuminating the target region and the background region respectively to obtain an updated pixel state set specifically includes: The pixels in the target area are illuminated by light of a preset first wavelength to improve the pixel state value of the target area. The pixels in the background area are illuminated by a preset second wavelength of light to reduce the pixel state value of the background area. Collect all pixel state values ​​after a preset illumination time to obtain an updated pixel state set.

[0008] According to some technical solutions of this application, the step of generating the current attention map based on the updated pixel state set specifically includes: A preset readout voltage is applied to the neural synaptic array; Collect the response current of each pixel under the read voltage in the updated pixel state set; The response current is converted into a digital signal to obtain a digital pixel group; The digitized pixel groups are arranged according to their pixel positions to generate the current attention map.

[0009] According to some technical solutions of this application, the initialization of the neural synapse array specifically includes: A third wavelength of light is applied to the neural synaptic array to reduce the state value of each pixel in the array; Read the state value of each pixel in the neural synapse array and determine whether the state value of each pixel reaches the preset baseline state range; If so, then the initialization of the neural synapse array is considered complete.

[0010] According to some technical solutions of this application, the step of outputting the target motion trajectory based on the current attention map specifically includes: Extract the current attention map to obtain the response peak region and its peak value; Determine whether the peak value is higher than the preset peak value threshold. If so, calculate the geometric center of the response peak value region and use the geometric center as the current target position. Obtain the historical trajectory sequence, and add the current target position to the end of the historical trajectory sequence to obtain the updated trajectory sequence; The updated trajectory sequence is output to obtain the target motion trajectory.

[0011] This application also provides a target motion trajectory output system, including: The initialization module is used to initialize the neural synapse array; The projection module is used to acquire optical images of the target scene and project the optical images onto the initialized neural synapse array to obtain an initial set of pixel states corresponding to the target scene. The processing module is used to process the initial pixel state set to determine the target area and the background area; The update module is used to illuminate the target area and the background area respectively to obtain an updated set of pixel states. The generation module is used to generate the current attention map based on the updated pixel state set; The output module is used to output the target's motion trajectory based on the current attention map.

[0012] This application also provides a target motion trajectory output device, the target motion trajectory output device comprising: a memory and at least one processor, wherein the memory stores instructions; At least one of the processors invokes the instructions in the memory to cause the target motion trajectory output device to execute the various steps of the target motion trajectory output method as described in the above technical solution.

[0013] This application also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the various steps of the target motion trajectory output method described above.

[0014] The target motion trajectory output method provided in this application has at least the following beneficial effects: the target motion trajectory is output through a fully optically controlled neural synapse array, and the state reinforcement of the target area and the state suppression of the background area are both completed by light, avoiding pixel-by-pixel electrical writing and complex gate control, making the system easier to make into an array and a low-power front end, and providing a relief path for the data transfer bottleneck of storage and computing separation, avoiding the repeated transfer of image data between the sensor, memory and processor in traditional solutions, thereby significantly reducing system power consumption and processing latency while maintaining high real-time performance. Attached Figure Description

[0015] Figure 1 A flowchart illustrating the target motion trajectory output method provided in this application embodiment; Figure 2 This is a schematic diagram of the target motion trajectory output system provided in the embodiments of this application; Figure 3 This is a schematic diagram of the target motion trajectory output device provided in an embodiment of this application. Detailed Implementation

[0016] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0017] In the description of this application, it should be understood that the orientation descriptions, such as up, down, front, back, left, right, etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation or be constructed or operated in a specific orientation. Therefore, they should not be construed as limiting the present invention.

[0018] In the description of this application, unless otherwise expressly defined, terms such as "setup," "installation," and "connection" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this application in conjunction with the specific content of the technical solution.

[0019] The following is combined Figures 1 to 3 Embodiments of the present invention will be described.

[0020] Traditional solutions suffer from high power consumption and difficulty in further reducing latency at the edge. They require a complete frame readout, transmission, and storage process before electronic inference at every moment. Much of the energy consumption does not come from the computation itself, but from the repeated data transfer between sensors, interfaces, memory, and processors. This bandwidth bottleneck and energy consumption caused by the separation of in-memory and computation is a fundamental problem that has long existed in the von Neumann architecture and is the main contradiction that near-in-memory computing and in-memory computing research have attempted to alleviate. Secondly, they are not robust enough to high-speed motion and complex lighting conditions. Frame-based imaging faces the contradiction of exposure time, noise, and motion blur. To reduce motion blur, exposure needs to be shortened, but this increases noise and reduces effective details; longer exposures cause ghosting of high-speed targets, rendering correlation matching and feature comparison ineffective. The third problem is that electronic domain remodeling models are difficult to achieve stable real-time performance at the edge. To improve robustness under occlusion, deformation, and background interference, many tracking systems are constantly increasing network size and computational complexity. This is feasible on desktop computing power, but it is prone to phenomena such as decreased inference frame rate, latency jitter, and thermal limitation on low-power devices, making it difficult to guarantee the "real-time performance and predictability of tracking output".

[0021] Based on the above, this application provides a method for outputting a target motion trajectory, including: S100, Initialize the neural synapse array; after system startup, 360nm ultraviolet light is used as the global suppression light to irradiate the fully optically controlled neural synapse array made of InGaZnO material for 2 seconds, causing the conductivity of all 128×128 pixels in the array to drop to the reference state of 0.1μS. Based on the device material characteristics test, InGaZnO has the highest absorption efficiency and the best suppression effect at a wavelength of 360nm, therefore it was selected as the suppression light wavelength.

[0022] S200 acquires optical images of the target scene and projects them onto an initialized neural synapse array to obtain an initial pixel state set corresponding to the target scene. Two frames of road scene images are continuously acquired through an imaging lens with a focal length of 50mm, and the optical images are projected onto the array surface. After the array receives light intensity, each pixel generates a change in conductivity corresponding to the light intensity, thus obtaining the initial pixel state set.

[0023] S300 processes the initial pixel state set to determine the target area and background area; reads the conductivity value of each pixel, determines the target area (i.e., the location of the vehicle) as the pixel with a conductivity value greater than 0.8 μS, and determines the background area (e.g., the road and sky) as the pixel with a conductivity value less than 0.2 μS.

[0024] S400: Irradiate the target area and the background area separately to obtain an updated pixel state set; for example, irradiate the target area with 460nm blue light for 0.2 seconds to increase the conductivity value, and irradiate the background area with 360nm ultraviolet light for 1 second to decrease the conductivity value, thus obtaining an updated pixel state set. Among these, the 460nm wavelength has the highest conductivity enhancement efficiency for InGaZnO, and the saturation gain can reach 15 times the initial value, so it is selected as the enhancement light wavelength.

[0025] S500 generates the current attention map based on the updated pixel state set; applies a 5V readout voltage to the array, collects the response current of each pixel, and generates a 256-level grayscale attention map.

[0026] The S600 outputs the target motion trajectory based on the current attention map. It calculates the centroid coordinates of the response peak region based on the attention map, and the coordinate sequence of 20 consecutive frames constitutes the vehicle motion trajectory, which is then output as the target motion trajectory. The system hardware includes an optical imaging lens, a 128×128 pixel fully optically controlled neural synapse array based on InGaZnO material, a micro-LED programmable light source array, readout circuitry, and an embedded processor.

[0027] In some embodiments, step S200 involves acquiring an optical image of the target scene and projecting the optical image onto an initialized neural synapse array to obtain an initial pixel state set corresponding to the target scene. Specifically, this includes: S210 continuously acquires at least two frames of optical images of the target scene to obtain an adjacent frame image set; the camera continuously acquires two frames of images at a frame rate of 50fps to obtain an adjacent frame image set.

[0028] S220, perform differential processing on adjacent frame image sets to obtain a differential image; perform pixel-level differential operation on two frames: subtract the corresponding pixel gray value of the first frame from the gray value of each pixel in the second frame image to obtain the differential image.

[0029] S230, extract the grayscale difference of each pixel in the difference image as the intensity change, and compare the intensity change of each pixel with a preset change threshold; for example, the pixel difference in row 50, column 30 is +45, and the pixel difference in row 60, column 80 is -38. Compare the absolute value of each difference with the preset change threshold of 30.

[0030] S240, if the intensity change is greater than the change threshold, it is determined to be a possible target region; for pixels with an absolute value greater than 30, such as "+45" and "-38", they are determined to be possible target regions, and a binary mask image is generated, where 1 represents a possible target region and 0 represents the background.

[0031] S250: Illuminate the potential target area and integrate the illuminated potential target area with the remaining unilluminated areas to obtain an initial pixel state set containing all pixel state values. Based on the mask image, project 460nm enhancement light for 0.2 seconds onto the pixel positions with a mask value of 1. After illumination, the conductivity of these pixels increases from a baseline of 0.1μS to 0.9μS; pixels with a mask value of 0 retain the baseline value of 0.1μS. Integrate the illuminated high-conductivity areas with the unilluminated low-conductivity areas to obtain an initial pixel state set of 128×128.

[0032] In some embodiments, step S300 processes the initial pixel state set to determine the target region and the background region, specifically including: S310, extract the state values ​​of each pixel in the initial pixel state set; read the current conductance values ​​of 128×128 pixels from the readout circuit of the neural synapse array to form the conductance matrix G[i][j].

[0033] S320: Compare the state value of each pixel in the initial pixel state set with a preset first threshold. If the state value is greater than the first threshold, the pixel is determined to be a target region. Compare each conductivity value G[i][j] with a preset first threshold of 0.8μS. If G[i][j] > 0.8μS, the pixel (i,j) is marked as a target region pixel. For example, 10 consecutive pixels satisfy the condition to form a strip-shaped region.

[0034] S330: Compare the state value of each pixel in the initial pixel state set with a preset second threshold. If the state value is less than the second threshold, the pixel is determined to be a background region. Compare each conductivity value G[i][j] with a preset second threshold of 0.2μS. If G[i][j] < 0.2μS, the pixel (i,j) is marked as a background region pixel. For example, large areas at the edge of the array satisfy this condition.

[0035] S340: Integrate the judgment results to obtain the target region and the background region. Denote the set of pixel coordinates of all pixels marked as the target region as T={(i,j)|G[i][j]>0.8μS}, and the set of pixel coordinates of all pixels marked as the background region as B={(i,j)|G[i][j]<0.2μS}. Output T and B as masks for subsequent illumination.

[0036] In some embodiments, step S400 involves illuminating the target region and the background region respectively to obtain an updated pixel state set, specifically including: S410: Illuminate the pixels in the target area with light of a preset first wavelength to improve the pixel state value of the target area; Based on the target area mask T obtained in S340, control the array composed of 1024 micro-LEDs to project blue light with a wavelength of 460nm onto the pixel position corresponding to T for 0.2 seconds, so that the conductivity value of these pixels is improved relative to the current value.

[0037] S420: Illuminate the pixels in the background area with light of a preset second wavelength to reduce the pixel state value of the background area; according to the background area mask B, project ultraviolet light with a wavelength of 360nm onto the pixel position corresponding to B for 1 second, so that the conductivity value of these pixels is reduced from the current value to close to the reference value of 0.1μS.

[0038] S430: Collect all pixel state values ​​after a preset illumination time to obtain an updated pixel state set. After the illumination is completed, collect the conductivity values ​​of all 128×128 pixels through the readout circuit to form an updated pixel state set, where the conductivity value of the target area is approximately 1.2 μS and the conductivity value of the background area is approximately 0.12 μS.

[0039] In some embodiments, S500, generating the current attention map based on the updated pixel state set, specifically includes: S510 applies a preset read voltage to the neural synapse array; for example, applying a 5V read voltage to the row lines of the fully optically controlled neural synapse array to put the array into read mode.

[0040] S520 collects the response current of each pixel under the read voltage in the updated pixel state set; through the current sensing amplifier on the column line, it collects the response current of each pixel under the 5V bias voltage row by row, with the current range from 0.1μA to 5μA, to obtain a 128×128 current value matrix I[i][j].

[0041] S530 converts the response current into a digital signal to obtain a digital pixel group; the analog current value is sent to a 12-bit analog-to-digital converter to be converted into a digital value of 0-4095 to obtain the digital pixel group D[i][j]. For example, the original current of 3.2μA is converted into the digital value 2621.

[0042] S540, arrange the digitized pixel group according to pixel position to generate the current attention map. Arrange the digitized pixel group D[i][j] according to the original pixel position to generate an 8-bit grayscale attention map of 512×512 pixels. The higher the brightness in the image, the greater the probability that the position is the target.

[0043] In some embodiments, S100, initializing the neural synapse array, specifically includes: S110, apply light of a third wavelength to the neural synapse array to reduce the state value of each pixel in the array; irradiate the entire 128×128 fully optically controlled neural synapse array with ultraviolet light of wavelength 360nm, light intensity 2mW / cm², for 2 seconds, so that the conductivity value of each pixel in the array gradually decreases from the original state of 0.5-2.0μS to the reference value of 0.1μS.

[0044] S120 reads the state values ​​of each pixel in the neural synapse array and determines whether the state value of each pixel reaches a preset reference state range. After irradiation, a 5V reading voltage is applied to the array through the readout circuit to collect the response current of all pixels and convert it into a conductance value. It then determines whether the conductance value of each pixel is within the reference state range of 0.09-0.11μS. For example, statistics show that the conductance value of the pixel in row 30, column 50 is 0.15μS, which is outside the range.

[0045] S130: If yes, the initialization of the neural synaptic array is considered complete. However, due to the presence of substandard pixels, return to S110 for another 1 second of illumination and repeat the reading. The second detection shows that the conductance values ​​of all pixels are within the range of 0.09-0.11 μS, indicating that initialization is complete, and the system enters S200 to prepare for tracking.

[0046] In some embodiments, step S600, based on the current attention map, outputs the target motion trajectory, specifically including: S610: Extract the current attention map to obtain the response peak region and its peak value; perform peak detection on the current attention map of 512×512 pixels to find the continuous region with the highest gray value, such as the rectangular region formed by rows 200-210 and columns 300-310, which has an average gray value of 245 and a peak gray value of 250.

[0047] S620: Determine if the peak value is higher than a preset peak value threshold. If so, calculate the geometric center of the response peak region and use the geometric center as the current target position. Determine if the peak value 250 is higher than the preset threshold of 200. Since 250 > 200, calculate the geometric center of the region: row center = (200 + 210) / 2 = 205.5, column center = (300 + 310) / 2 = 305, and round down to obtain the current target position (206, 305).

[0048] S630, Obtain the historical trajectory sequence, and add the current target position to the end of the historical trajectory sequence to obtain the updated trajectory sequence; Obtain the historical trajectory sequence L stored in memory, currently recording the positions of the previous 19 frames [(201,300), (202,301), ... , (205,304)]. Add the current target position (206,305) to the end of the sequence to obtain the updated trajectory sequence L' containing 20 coordinate points.

[0049] S640 outputs the updated trajectory sequence to obtain the target motion trajectory. The updated trajectory sequence L' is then output to plot the continuous motion trajectory of the vehicle.

[0050] This application also provides a target motion trajectory output system, including: an initialization module 100, a projection module 200, a processing module 300, an update module 400, a generation module 500, and an output module 600.

[0051] The initialization module 100 is used to initialize the neural synapse array; the projection module 200 is used to acquire optical images of the target scene and project the optical images onto the initialized neural synapse array to obtain an initial pixel state set corresponding to the target scene; the processing module 300 is used to process the initial pixel state set to determine the target region and the background region; the update module 400 is used to illuminate the target region and the background region respectively to obtain an updated pixel state set; the generation module 500 is used to generate a current attention map based on the updated pixel state set; and the output module 600 is used to output the target motion trajectory based on the current attention map.

[0052] Figure 3 This is a schematic diagram of a target motion trajectory output device 700 provided in an embodiment of the present invention. The target motion trajectory output device 700 can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 710 (e.g., one or more processors) and a memory 720, and one or more storage media 730 (e.g., one or more mass storage devices) storing application programs 733 or data 732. The memory 720 and storage media 730 can be temporary or persistent storage. The program stored in the storage media 730 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the target motion trajectory output device 700. Furthermore, the processor 710 may be configured to communicate with the storage media 730 and execute the series of instruction operations in the storage media 730 on the target motion trajectory output device 700 to implement the steps of the target motion trajectory output method provided in the above-described method embodiments.

[0053] The target motion trajectory output device 700 may also include one or more power supplies 740, one or more wired or wireless network interfaces 750, one or more input / output interfaces 760, and / or one or more operating systems 731, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3 The structure of the target motion trajectory output device shown does not constitute a limitation on the target motion trajectory output device. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0054] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of a method for outputting a target motion trajectory.

[0055] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0056] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0057] The preferred embodiments of the present invention have been described in detail above, but the present disclosure is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention, and these equivalent modifications or substitutions are all included within the scope defined by the claims of the present disclosure.

Claims

1. A method for outputting the trajectory of a target motion, characterized in that: include: Initialize the neural synapse array; Acquire optical images of the target scene and project them onto an initialized neural synapse array to obtain an initial set of pixel states corresponding to the target scene; The initial pixel state set is processed to determine the target region and the background region; The target region and the background region are illuminated separately to obtain an updated pixel state set; Generate the current attention map based on the updated pixel state set; Based on the current attention map, output the target's motion trajectory; The step of illuminating the target region and the background region separately to obtain the updated pixel state set specifically includes: The pixels in the target area are illuminated by light of a preset first wavelength to improve the pixel state value of the target area. The pixels in the background area are illuminated by a preset second wavelength of light to reduce the pixel state value of the background area. Collect all pixel state values ​​after a preset illumination time to obtain an updated pixel state set.

2. The target motion trajectory output method according to claim 1, characterized in that: The process of acquiring optical images of the target scene and projecting these images onto an initialized neural synapse array to obtain an initial set of pixel states corresponding to the target scene includes: At least two frames of optical images of the target scene are continuously acquired to obtain a set of adjacent frame images; Differential processing is performed on adjacent frame image sets to obtain differential images; Extract the grayscale difference of each pixel in the difference image as the intensity change, and compare the intensity change of each pixel with a preset change threshold; If the intensity change is greater than the change threshold, it is identified as a possible target area; The potential target area is illuminated, and the illuminated potential target area is integrated with the remaining unilluminated areas to obtain an initial pixel state set containing all pixel state values.

3. The target motion trajectory output method according to claim 1, characterized in that: The process of processing the initial pixel state set to determine the target region and the background region specifically includes: Extract the state values ​​of each pixel from the initial pixel state set; The state value of each pixel in the initial pixel state set is compared with a preset first threshold. If the state value is greater than the first threshold, the pixel is determined to be the target region. The state value of each pixel in the initial pixel state set is compared with a preset second threshold. If the state value is less than the second threshold, the pixel is determined to be a background area. The judgment results are integrated to obtain the target area and the background area.

4. The target motion trajectory output method according to claim 1, characterized in that: The step of generating the current attention map based on the updated pixel state set specifically includes: A preset readout voltage is applied to the neural synaptic array; Collect the response current of each pixel under the read voltage in the updated pixel state set; The response current is converted into a digital signal to obtain a digital pixel group; The digitized pixel groups are arranged according to their pixel positions to generate the current attention map.

5. The target motion trajectory output method according to claim 1, characterized in that: The initialization of the neural synapse array specifically includes: A third wavelength of light is applied to the neural synaptic array to reduce the state value of each pixel in the array; Read the state value of each pixel in the neural synapse array and determine whether the state value of each pixel reaches the preset baseline state range; If so, then the initialization of the neural synapse array is considered complete.

6. The target motion trajectory output method according to claim 1, characterized in that: The step of outputting the target motion trajectory based on the current attention map specifically includes: Extract the current attention map to obtain the response peak region and its peak value; Determine whether the peak value is higher than the preset peak value threshold. If so, calculate the geometric center of the response peak value region and use the geometric center as the current target position. Obtain the historical trajectory sequence, and add the current target position to the end of the historical trajectory sequence to obtain the updated trajectory sequence; The updated trajectory sequence is output to obtain the target motion trajectory.

7. A target motion trajectory output system, characterized in that: include: The initialization module is used to initialize the neural synapse array; The projection module is used to acquire optical images of the target scene and project the optical images onto the initialized neural synapse array to obtain an initial set of pixel states corresponding to the target scene. The processing module is used to process the initial pixel state set to determine the target area and the background area; The update module is used to illuminate the target area and the background area respectively to obtain an updated set of pixel states. The step of illuminating the target area and the background area respectively to obtain the updated pixel state set specifically includes: illuminating the pixels in the target area with light of a preset first wavelength to improve the pixel state value of the target area; and illuminating the pixels in the background area with light of a preset second wavelength to reduce the pixel state value of the background area. Collect all pixel state values ​​after a preset illumination time to obtain an updated pixel state set; The generation module is used to generate the current attention map based on the updated pixel state set; The output module is used to output the target's motion trajectory based on the current attention map.

8. A target motion trajectory output device, characterized in that: The target motion trajectory output device includes: a memory and at least one processor, wherein the memory stores instructions; At least one of the processors invokes the instructions in the memory to cause the target motion trajectory output device to perform the steps of the target motion trajectory output method as described in any one of claims 1-6.

9. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the various steps of the target motion trajectory output method as described in any one of claims 1-6.

Images