Target edge detection and recognition method, device, equipment and medium under single pixel system

By fusing the Sobel edge detection operator with the Hard code light field matrix in single pixel imaging technology to generate an edge detection structured light field, combining gradient detection and Fourier descriptor, it solves the difficulty of obtaining edge information and real-time in single pixel imaging. The problem of poor performance achieves efficient target profile feature extraction and recognition.

CN117911711BActive Publication Date: 2024-11-12NAT UNIV OF DEFENSE TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311796738.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-12-25
Publication Date
2024-11-12
Estimated Expiration
2043-12-25

AI Technical Summary

Technical Problem

In single-pixel imaging technology, it is difficult to acquire edge information of the target and poor real-time performance. The background noise caused by the structured light field characteristics reduces the robustness of the edge detection algorithm.

Method used

The Sobel edge detection operator is fused with the two-dimensional intensity distribution matrix of the Hard code light field to generate an edge detection structured light field. The target edge profile is extracted through gradient detection, and the target edge features are characterized by Fourier descriptors to achieve robustness. Good image features.

Benefits of technology

The speed of single-pixel imaging system extracting and identifying target profile features has been improved, and the practicalization of single-pixel imaging technology in the field of long-distance target laser detection has been promoted.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117911711B_ABST
    Figure CN117911711B_ABST
Patent Text Reader

Abstract

The present invention proposes a method, device, equipment and medium for target edge detection and recognition under a single pixel system, including: obtaining the echo intensity corresponding to the target when the edge detection structured light field illuminates the target, the edge detection structured light field is modulated by the edge detection structured light field two-dimensional intensity division matrix, and the edge detection structured light field two-dimensional intensity division matrix is ​​obtained by fusing the Sobel edge detection operator with the Hader code matrix; reconstructing the target image based on the echo intensity; obtaining the target gradient map based on the target image; performing gradient detection based on the target gradient map to extract the target edge contour, and then extracting the Fourier descriptor feature vector according to the target contour curve to classify the target. The present invention realizes the edge detection of the target image by the single pixel imaging system, improves the speed of target contour feature extraction, and helps to further promote the practical application of single pixel imaging technology in the field of long-distance target laser detection.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention mainly relates to the field of single-pixel imaging technology, and in particular to a method, device, equipment and medium for detecting and identifying target edges under a single-pixel system. Background Art

[0002] Edge detection technology is an important means to obtain image structural features. Edge detection can significantly reduce the amount of image data, eliminate irrelevant information, and retain the inherent characteristics of the target image, so it is of great significance in image recognition.

[0003] In a single-pixel imaging system, since a single-pixel detector only receives the intensity signal of the target echo, it is very difficult to decode complex target information from the target echo signal. Therefore, the current mainstream single-pixel imaging technology mainly focuses on high-quality and fast imaging of the target. If you want to obtain the edge information of the target, you can only extract the edge of the target based on the clear image after calculating the imaging of the target.

[0004] Since single-pixel imaging requires a large number of structured light fields to illuminate the target, the imaging speed under the same conditions is not as fast as that of traditional array sensor imaging. Therefore, it is difficult to obtain high real-time performance for edge detection of the target under the single-pixel imaging system. In addition, the academic community has proven that due to the characteristics of certain structured light fields, the target images obtained by single-pixel imaging technology often contain some background noise, which reduces the robustness of traditional edge detection algorithms. Summary of the invention

[0005] In view of the technical problems existing in the prior art, the present invention proposes a method, device, equipment and medium for target edge detection and recognition under a single-pixel system.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0007] On the one hand, the present invention provides a method for detecting an edge of an object in a single pixel system, comprising:

[0008] Acquire the echo intensity corresponding to the target when the edge detection structured light field is illuminated; wherein the edge detection structured light field is modulated by a two-dimensional intensity division matrix of the edge detection structured light field, and the two-dimensional intensity division matrix of the edge detection structured light field is obtained by fusing the Sobel edge detection operator with the two-dimensional intensity distribution matrix of the Hadyard light field;

[0009] reconstructing a target image based on the echo intensity;

[0010] Obtaining a gradient map of the target based on the target image;

[0011] The target edge contour can be extracted by performing gradient detection based on the target's gradient map.

[0012] Furthermore, the method for generating the two-dimensional intensity division matrix of the edge detection structured light field includes:

[0013] Generate a two-dimensional intensity distribution matrix of the Haddon light field with a size of N×N;

[0014] The size of each group of Hardcode light field two-dimensional intensity distribution matrix is ​​enlarged by 3 times;

[0015] The pixel blocks with a scale of 3×3 and values ​​of all +1 on each enlarged Hardy light field two-dimensional intensity distribution matrix are replaced with the Sobel edge detection operator to obtain the edge detection structured light field two-dimensional intensity distribution matrix.

[0016] Furthermore, the Sobel edge detection operator includes S x and S y , S x and S y Represent the edge detection operators in the x and y directions respectively, where:

[0017] The present invention provides a method for generating a two-dimensional intensity distribution matrix of an edge detection structured light field, which is obtained by fusing a Sobel edge detection operator with a two-dimensional intensity distribution matrix of a Hadley code light field. Specifically, the method includes:

[0018] Generate a two-dimensional intensity distribution matrix of the Haddon light field with a size of N×N;

[0019] The size of each group of Hardcode light field two-dimensional intensity distribution matrix is ​​enlarged by 3 times;

[0020] Each pixel block with a scale of 3×3 and a value of +1 on the enlarged two-dimensional intensity distribution matrix of the Haddon light field is replaced with a Sobel edge detection operator to obtain a two-dimensional intensity distribution matrix of the edge detection structured light field.

[0021] The present invention provides a method for generating an edge detection structured light field, comprising:

[0022] Generate a two-dimensional intensity distribution matrix of the edge detection structured light field, wherein the two-dimensional intensity distribution matrix of the edge detection structured light field is obtained by fusing a Sobel edge detection operator with a two-dimensional intensity distribution matrix of the Hadcode light field;

[0023] The edge detection structured light field is obtained by modulating the illumination beam using a two-dimensional intensity division matrix.

[0024] On the other hand, the present invention also provides a method for obtaining target edge contour features, comprising:

[0025] Extracting the target edge contour of the target based on the target edge detection method under the above single pixel system;

[0026] The Fourier descriptor a(u) is used to characterize the edge contour features of the target;

[0027]

[0028] Where k = 0, 1, 2...K-1. K is the number of sampling points of discrete Fourier transform, and u is the signal frequency.

[0029] The Fourier descriptor only needs a small number of descriptor items (that is, the number in the descriptor vector does not need to be too large) to roughly represent the entire target edge contour. After a simple normalization operation on the Fourier descriptor, the Fourier descriptor can be made translation, rotation, and scale invariant, that is, it is not affected by the position, angle, and scaling of the target edge contour in the target image, and is a robust image feature.

[0030] On the other hand, the present invention also provides a target recognition method based on target edge contour, comprising:

[0031] Extracting the target edge contour of the target based on the target edge detection method under the above single pixel system;

[0032] The Fourier descriptor a(u) is used to characterize the edge contour features of each target;

[0033]

[0034] Where k = 0, 1, 2...K-1. K is the order of discrete Fourier transform. K is the number of sampling points of discrete Fourier transform, and u is the signal frequency.

[0035] Based on the Fourier descriptor a(u), the edge contours of different targets are identified and distinguished, and then different targets are distinguished. After the target edge contour of the target is extracted, the target edge contour features are characterized by the Fourier descriptor, and then different targets are identified according to the Fourier descriptor, and then different targets can be classified. This method can realize target contour recognition without relying on the target image in a single-pixel imaging system.

[0036] On the other hand, the present invention provides a single-pixel target edge detection device, comprising:

[0037] An edge detection structured light field two-dimensional intensity distribution matrix generating unit is used to generate an edge detection structured light field two-dimensional intensity distribution matrix, wherein the edge detection structured light field two-dimensional intensity distribution matrix is ​​obtained by fusing a Sobel edge detection operator with a Hadley light field two-dimensional intensity distribution matrix;

[0038] An illumination beam generating unit, comprising a light source and a light field modulator, the light field modulator being loaded with the edge detection structure light field two-dimensional intensity division matrix, and being used for modulating the light beam generated by the light source, and irradiating the modulated light beam onto the target;

[0039] A target echo acquisition unit, used to obtain the echo intensity corresponding to the target when the edge detection structured light field illuminates the target;

[0040] The target edge contour extraction unit is used to reconstruct the target image based on the echo intensity, obtain the target gradient map based on the target image, and perform gradient detection based on the target gradient map to extract the target edge contour.

[0041] In another aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0042] Acquire the echo intensity corresponding to the target when the edge detection structured light field illuminates the target, the number of targets being 1 or more; wherein the edge detection structured light field is modulated by a two-dimensional intensity division matrix of the edge detection structured light field, and the two-dimensional intensity division matrix of the edge detection structured light field is obtained by fusing a Sobel edge detection operator with a two-dimensional intensity distribution matrix of a Hadyard light field;

[0043] reconstructing a target image based on the echo intensity;

[0044] Obtaining a gradient map of the target based on the target image;

[0045] The target edge contour can be extracted by performing gradient detection based on the target's gradient map.

[0046] In another aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the following steps:

[0047] Acquire the echo intensity corresponding to the target when the edge detection structured light field illuminates the target, the number of targets being 1 or more; wherein the edge detection structured light field is modulated by a two-dimensional intensity division matrix of the edge detection structured light field, and the two-dimensional intensity division matrix of the edge detection structured light field is obtained by fusing a Sobel edge detection operator with a two-dimensional intensity distribution matrix of a Hadyard light field;

[0048] reconstructing a target image based on the echo intensity;

[0049] Obtaining a gradient map of the target based on the target image;

[0050] The target edge contour can be extracted by performing gradient detection based on the target's gradient map.

[0051] Compared with the prior art, the technical effects of the present invention are:

[0052] The present invention provides an edge detection technology solution under the single-pixel imaging technology system. By fusing the Sobel edge detection operator with the Hader code matrix, a new edge detection structured light field is designed. When the target is illuminated by the light field, the target contour can be directly obtained by using the traditional gradient difference (GI) single-pixel image reconstruction algorithm, and then the target is classified according to the Fourier descriptor feature vector extracted according to the target contour curve. The present invention realizes the edge detection of the target image by the single-pixel imaging system, improves the speed of target contour feature extraction and recognition, and helps to further promote the practical application of single-pixel imaging technology in the field of long-distance target laser detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0054] Figure 1 is a flow chart of a method for detecting an edge of an object in a single pixel system provided by an embodiment;

[0055] Figure 2 It is a schematic diagram of generating a two-dimensional distribution matrix of the intensity of a structured light field for edge detection according to an embodiment;

[0056] Figure 3 is a structural diagram of an object edge detection device in a single pixel system provided by an embodiment;

[0057] Figure 4 This is a diagram showing the effect of edge extraction of a target using a target edge detection method under a single pixel system in an embodiment, wherein (a) is the target and (b) is the edge contour of the target;

[0058] Figure 5 1 is a target classification effect of a target edge detection method using a single-pixel system in an embodiment, wherein (a) is the Fourier descriptor of target 1 and the target classification effect in different postures, wherein target 1 is a drone, and (b) is the Fourier descriptor of target 2 and the target classification effect in different postures, wherein target 2 is a fixed-wing aircraft. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] In one embodiment of the present invention, referring to Figure 1 , a target edge detection method under a single pixel system is provided, comprising:

[0061] Obtain the echo intensity corresponding to the target illuminated by the edge detection structured light field;

[0062] reconstructing a target image based on the echo intensity;

[0063] Obtaining a gradient map of the target based on the target image;

[0064] The target edge contour can be extracted by performing gradient detection based on the target's gradient map.

[0065] Since the gradient of the target edge contour pixels is generally higher than the gradient of the other pixels in the target image, the target edge contour can be extracted by performing gradient detection on the target gradient map.

[0066] The number of targets in the present invention is not limited, the number of targets can be 1 or more, and the contour lines of multiple target images can be obtained at the same time; wherein the edge detection structured light field is modulated by the edge detection structured light field two-dimensional intensity division matrix, and the edge detection structured light field two-dimensional intensity division matrix is ​​obtained by fusing the Sobel edge detection operator with the Hadcode light field two-dimensional intensity distribution matrix.

[0067] In single-pixel imaging illumination, a matrix sequence representing the two-dimensional distribution of light field intensity is generally used to represent a structured light field. In one embodiment, the method for generating the two-dimensional intensity distribution matrix of the edge detection structured light field includes:

[0068] Generate a two-dimensional intensity distribution matrix of the Haddon light field with a size of N×N;

[0069] The size of each group of Hardcode light field two-dimensional intensity distribution matrix is ​​enlarged by 3 times;

[0070] The pixel blocks with a scale of 3×3 and values ​​of all +1 on each enlarged Hardy light field two-dimensional intensity distribution matrix are replaced with the Sobel edge detection operator to obtain the edge detection structured light field two-dimensional intensity distribution matrix.

[0071] The Sobel edge detection operator includes S x and S y , Sx and S y Represent the edge detection operators in the x and y directions respectively, where:

[0072] Reference Figure 2 , Figure 2 This is a schematic diagram of generating a two-dimensional intensity distribution matrix of a structured light field for edge detection according to an embodiment. First, a two-dimensional intensity distribution matrix of a Haddon light field with a size of N×N is generated, and the intensity distribution matrix is ​​enlarged by 3 times to 3N×3N. Figure 2 On the matrix, a white square has a size of 3×3 pixels and a pixel value of 1. Replace the pixel blocks with a scale of 3×3 and a value of +1 with Sobel operators in the x and y directions respectively, and you can get the two-dimensional intensity distribution matrix of the edge detection structured light field in the x and y directions.

[0073] Further, the edge detection structured light field includes edge detection structured light fields in the x direction and the y direction;

[0074] Based on the target image, the gradient image G of the target in the x and y directions is obtained. x and G y ;

[0075] Get the target's gradient map based on the target's gradient image in the x and y directions

[0076] For single-pixel imaging systems, edge detection operators are not orthogonal and are too small in scale, so they cannot be directly used in single-pixel imaging systems. By fusing edge detection operators with orthogonal structured light fields such as Hardy code, a new type of structured light field with edge detection capability is obtained, and then image reconstruction algorithms such as GI can realize contour detection of the target image.

[0077] In one embodiment, a new edge detection structured light field two-dimensional intensity distribution matrix is ​​designed by fusing the Sobel edge detection operator with the Hadcode light field two-dimensional intensity distribution matrix. Specifically, in one embodiment, a method for generating a two-dimensional intensity distribution matrix of an edge detection structured light field is provided, which is obtained by fusing the Sobel edge detection operator with the Hadcode light field two-dimensional intensity distribution matrix. Specifically, it includes:

[0078] Generate a two-dimensional intensity distribution matrix of the Haddon light field with a size of N×N;

[0079] The size of each group of Hardcode light field two-dimensional intensity distribution matrix is ​​enlarged by 3 times;

[0080] Each pixel block with a scale of 3×3 and a value of +1 on the enlarged two-dimensional intensity distribution matrix of the Haddon light field is replaced with a Sobel edge detection operator to obtain a two-dimensional intensity distribution matrix of the edge detection structured light field.

[0081] In one embodiment, a new edge detection structured light field is designed by fusing the Sobel edge detection operator with the two-dimensional intensity distribution matrix of the Haddon light field. The structured light field has a stronger response to the edge information of the target and can be directly used for edge detection of the target. Specifically, a method for generating an edge detection structured light field includes:

[0082] Generate a two-dimensional intensity distribution matrix of the edge detection structured light field, wherein the two-dimensional intensity distribution matrix of the edge detection structured light field is obtained by fusing a Sobel edge detection operator with a two-dimensional intensity distribution matrix of the Hadcode light field;

[0083] The edge detection structured light field is obtained by modulating the illumination light beam using the edge detection structured light field two-dimensional intensity division matrix. The specific implementation method is not limited, such as loading the edge detection structured light field two-dimensional intensity division matrix onto a light field modulation device such as a DMD, and the illumination light beam is modulated to obtain the edge detection structured light field after passing through the light field modulation device loaded with the edge detection structured light field two-dimensional intensity division matrix.

[0084] When the target is illuminated by the above-mentioned edge detection structured light field, the target contour curve can be directly obtained by using the traditional gradient difference (GI) single pixel image reconstruction algorithm, and then the target is classified by extracting the Fourier descriptor feature vector according to the target contour curve. Therefore, the present invention can realize target edge contour detection and target recognition without relying on target imaging under a single pixel imaging system.

[0085] In one embodiment, a method for obtaining edge contour features of a target is provided, comprising:

[0086] (1) Extracting the target edge contour of the target, including:

[0087] Obtain the echo intensity corresponding to the target illuminated by the edge detection structured light field;

[0088] reconstructing a target image based on the echo intensity;

[0089] Obtaining a gradient map of the target based on the target image;

[0090] The target edge contour can be extracted by performing gradient detection based on the target's gradient map.

[0091] Since the gradient of the target edge contour pixels is generally higher than the gradient of the other pixels in the target image, the target edge contour can be extracted by performing gradient detection on the target gradient map.

[0092] (2) Use Fourier descriptor a(u) to characterize the edge contour features of the target;

[0093]

[0094] Where k = 0, 1, 2...K-1, K is the number of sampling points of discrete Fourier transform, u is the signal frequency, and s(k) represents the edge contour of the target.

[0095] The method for generating the edge detection structured light field in the above-mentioned embodiment has been described in detail in the aforementioned embodiment and will not be repeated here. When edge contour detection is performed on a target, the number of targets is not limited, and can be one or more. Therefore, the present invention has the potential to extract the centroid of multiple targets at the same time. The Fourier descriptor is extracted to characterize the edge contour features of the target. The basic idea is to use the Fourier transform of the object boundary information as the shape feature, transform the contour feature from the spatial domain to the frequency domain, and extract the frequency domain information as the feature vector of the image. That is, a vector is used to represent a contour, and the contour is digitized, so that different contours can be better distinguished. The Fourier descriptor has the characteristics of rotation, translation and scale invariance, and can be widely used in target recognition.

[0096] On the other hand, the present invention also provides a target recognition method based on target edge contour, comprising:

[0097] (1) Extracting the target edge contour of the target, including:

[0098] Obtain the echo intensity corresponding to the target illuminated by the edge detection structured light field;

[0099] reconstructing a target image based on the echo intensity;

[0100] The target edge contour can be extracted by performing gradient detection based on the target's gradient map.

[0101] The target gradient map is obtained based on the target image; the target gradient map contains the edge information of the target image. The target edge contour can be obtained by binarizing the target gradient map. Since the gradient of the target edge contour pixels is generally higher than the gradient of the remaining pixels in the target image, the target edge contour can be extracted by performing gradient detection on the target gradient map.

[0102] (2) Use Fourier descriptor a(u) to characterize the edge contour features of each target;

[0103]

[0104] Where k = 0, 1, 2...K-1, K is the number of sampling points of discrete Fourier transform, and u is the signal frequency.

[0105] (3) Based on the Fourier descriptor a(u), the edge contours of different targets are identified and distinguished, thereby distinguishing different targets.

[0106] After extracting the target edge contour, the target edge contour features are characterized by Fourier descriptors, and then different targets are identified according to the Fourier descriptors, and different targets can also be classified. This method can realize target contour recognition without relying on the target image under a single-pixel imaging system. The Fourier descriptor requires a small number of descriptor items (that is, the number in the descriptor vector does not need to be too large) to roughly represent the entire target edge contour. After a simple normalization operation on the Fourier descriptor, the Fourier descriptor can be made translation, rotation, and scale invariant, that is, it is not affected by the position, angle, and scaling of the target edge contour in the image, and is a robust image feature.

[0107] In another embodiment, a single-pixel target edge detection device is provided, comprising:

[0108] An edge detection structured light field two-dimensional intensity distribution matrix generating unit is used to generate an edge detection structured light field two-dimensional intensity distribution matrix, wherein the edge detection structured light field two-dimensional intensity distribution matrix is ​​obtained by fusing a Sobel edge detection operator with a Hadley light field two-dimensional intensity distribution matrix;

[0109] An illumination beam generating unit, comprising a light source and a light field modulator, the light field modulator being loaded with the edge detection structure light field two-dimensional intensity division matrix, and being used for modulating the light beam generated by the light source, and irradiating the modulated light beam onto the target;

[0110] A target echo acquisition unit, used to obtain the echo intensity corresponding to the target when the edge detection structured light field illuminates the target;

[0111] The target edge contour extraction unit is used to reconstruct the target image based on the echo intensity, obtain the target gradient map based on the target image, and perform gradient detection based on the target gradient map to extract the target edge contour.

[0112] The edge detection structured light field two-dimensional intensity indexing matrix generating unit generates an edge detection structured light field two-dimensional intensity indexing matrix, wherein the edge detection structured light field two-dimensional intensity indexing matrix is ​​obtained by fusing the Sobel edge detection operator with the Hadcode light field two-dimensional intensity distribution matrix, specifically, including:

[0113] Generate a two-dimensional intensity distribution matrix of the Haddon light field with a size of N×N;

[0114] The size of each group of Hardcode light field two-dimensional intensity distribution matrix is ​​enlarged by 3 times;

[0115] The pixel blocks with a scale of 3×3 and all values ​​of +1 on each enlarged Hardy light field two-dimensional intensity distribution matrix are replaced with Sobel edge detection operators in the x-direction and y-direction respectively to obtain the edge detection structured light field two-dimensional intensity distribution matrix in the x-direction and y-direction.

[0116] The two-dimensional intensity division matrix of the edge detection structured light field in the x-direction and the y-direction is loaded onto a light field modulation device such as a DMD. After the illumination light beam passes through the light field modulation device loaded with the two-dimensional intensity division matrix of the edge detection structured light field, it is modulated to obtain the edge detection structured light field in the x-direction and the y-direction.

[0117] The edge detection structured light field in the x-direction and y-direction is used to illuminate the target, and a single-pixel detector is used to record the corresponding echo intensity.

[0118] Based on the echo intensity, the target image is calculated using the traditional GI single-pixel image reconstruction algorithm to obtain the gradient image G of the target in the x and y directions. x and G y ; Get the target's gradient map based on the target's gradient image in the x and y directions Since in the target's gradient map, the pixel value of the target edge is the gradient value corresponding to the target image pixel, the target's gradient map is binarized, and the gradient of the target edge contour pixel is higher than the gradient of the remaining pixels in the target image, and the target edge contour curve can be obtained.

[0119] In another embodiment, a single-pixel target recognition device is provided, comprising:

[0120] The target edge detection device in a single-pixel system described in any of the above embodiments is used to extract the edge contour of the target;

[0121] A target edge contour feature extraction unit, used to characterize the edge contour features of each target using Fourier descriptor a(u);

[0122] The recognition unit recognizes and distinguishes the edge contours of different targets based on the Fourier descriptor a(u), thereby distinguishing different targets.

[0123] Figure 3 It is a structural diagram of a target edge detection device in a single-pixel system provided by an embodiment. Figure 3In the invention, two DMD spatial light modulators are used, one of which is used to modulate the incident laser beam, and the other is used to display the target image. The illumination laser emitted by the laser 1 is expanded by the beam expander 2 and then hits the first DMD spatial light modulator 3. The first DMD spatial light modulator 3 is loaded with the edge detection light field two-dimensional intensity division matrix, thereby modulating the illumination laser. The modulated laser beam is irradiated to the second DMD spatial light modulator 5 that displays the target through the emission lens 4. Since the target is displayed by the second DMD spatial light modulator 5, it is convenient to rotate, scale and translate the target, so that the effectiveness of the method of the present invention can be verified. The return light from the target is incident on the single pixel detector 7 after passing through the focusing lens 6. The single pixel detector 7 is used to record the corresponding echo intensity, and the output of the single pixel detector 7 is sampled by the sampling unit and transmitted to the computer 9. The edge detection light field two-dimensional intensity division matrix is ​​generated by the computer and loaded onto the first DMD spatial light modulator 3. At the same time, the computer 9 extracts the target contour curve, extracts the target contour features, and performs target recognition.

[0124] Figure 4 FIG. 1 is a diagram showing the effect of edge extraction of a target using a target edge detection method under a single pixel system in an embodiment, wherein (a) is the target and (b) is the target edge contour line; Figure 4 , the present invention adopts the GI single pixel image reconstruction algorithm to obtain the target edge contour line instead of the target image, which illustrates the effectiveness of the method of the present invention.

[0125] Figure 5 This is the target classification effect of a target edge detection method using a single pixel system in one embodiment. Figure 5 In the figure, the method of the present invention is used to detect the edges of two targets, and 16 Fourier descriptors are extracted respectively. Among them, (a) is the Fourier descriptor of target 1 and the target classification effect under different postures, where target 1 is a drone, and (b) is the Fourier descriptor of target 2 and the target classification effect under different postures, where target 2 is a fixed-wing aircraft. According to Figure 5 The rotation, scaling and translation of the target itself have little effect on the Fourier descriptor. The Fourier descriptor vectors of the same target at different angles, scales and translations are basically the same. At the same time, according to the Fourier descriptor, it is easy to classify drones and fixed-wing aircraft targets. Furthermore, when the leaf descriptor vector of the target is known, the target can be effectively and accurately identified.

[0126] On the other hand, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the target edge detection method under the single-pixel system provided in any of the above embodiments are implemented. The computer device can be a server. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store sample data. The network interface of the computer device is used to communicate with an external terminal through a network connection.

[0127] On the other hand, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the target edge detection method in a single-pixel system provided in any of the above embodiments.

[0128] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0129] Matters not covered by the present invention are known technologies.

[0130] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0131] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

[0132] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A target edge detection method under a single pixel system, characterized in that: include: Obtain the echo intensity corresponding to the target illuminated by the edge detection structured light field, where the number of targets is 1 or more; The edge detection structured light field is obtained by modulating a two-dimensional intensity division matrix of the edge detection structured light field, and the two-dimensional intensity division matrix of the edge detection structured light field is obtained by fusing a Sobel edge detection operator with a two-dimensional intensity distribution matrix of a Hadley light field. The method for generating the two-dimensional intensity division matrix of the edge detection structured light field includes: Generate a two-dimensional intensity distribution matrix of the Haddon light field with a size of N×N; The size of each group of Hardcode light field two-dimensional intensity distribution matrix is ​​enlarged by 3 times; The pixel blocks with a scale of 3×3 and values ​​of all +1 on each enlarged two-dimensional intensity distribution matrix of the Haddon light field are replaced with the Sobel edge detection operator to obtain the two-dimensional intensity distribution matrix of the edge detection structured light field; reconstructing a target image based on the echo intensity; Obtaining a gradient map of the target based on the target image; The target edge contour can be extracted by performing gradient detection based on the target's gradient map.

2. The target edge detection method under the single pixel system according to claim 1 is characterized in that: The echo intensity is used to reconstruct the target image through the GI single pixel image reconstruction method.

3. The target edge detection method under the single pixel system according to claim 1 is characterized in that: The Sobel edge detection operator includes S x and S y , S x and S y Represent the edge detection operators in the x and y directions respectively, where: .

4. The object edge detection method in a single pixel system according to any one of claims 1 to 3, characterized in that: The edge detection structured light field includes edge detection structured light fields in the x-direction and the y-direction; The gradient map of the target obtained based on the target image includes: Based on the target image, the gradient image of the target in the x and y directions is obtained G x and G y ; Get the target's gradient map based on the target's gradient image in the x and y directions .

5. The method for generating a structured light field for edge detection is characterized in that: include: Generate a two-dimensional intensity indexing matrix of an edge detection structured light field, wherein the two-dimensional intensity indexing matrix of an edge detection structured light field is obtained by fusing a Sobel edge detection operator with a two-dimensional intensity distribution matrix of a Hadley light field. The method for generating the two-dimensional intensity indexing matrix of an edge detection structured light field comprises: Generate a two-dimensional intensity distribution matrix of the Haddon light field with a size of N×N; The size of each group of Hardcode light field two-dimensional intensity distribution matrix is ​​enlarged by 3 times; The pixel blocks with a scale of 3×3 and values ​​of all +1 on each enlarged two-dimensional intensity distribution matrix of the Haddon light field are replaced with the Sobel edge detection operator to obtain the two-dimensional intensity distribution matrix of the edge detection structured light field; The edge detection structured light field is obtained by modulating the illumination beam using a two-dimensional intensity division matrix.

6. The method for generating an edge detection structured light field according to claim 5, characterized in that: The Sobel edge detection operator includes S x and S y , S x and S y Represent the edge detection operators in the x and y directions respectively, where: .

7. A method for obtaining target edge contour features, characterized in that: Extract the target edge contour of the target based on the target edge detection method under the single pixel system as claimed in claim 1 ; Fourier descriptor Characterize the edge contour features of the target; in k=0,1,2...K -1, K is the number of sampling points of discrete Fourier transform, u is the signal frequency.

8. A target recognition method based on target edge contour, characterized in that: include: Extract target edge contours of different targets based on the target edge detection method under the single pixel system as claimed in claim 1 ; Fourier descriptor Characterize the edge contour features of each target; in k=0,1,2...K -1, K is the number of sampling points of discrete Fourier transform, u is the signal frequency; Based on Fourier descriptor Identify and distinguish the edge contours of different targets, and then distinguish different targets.

9. A target edge detection device in a single pixel system, characterized in that: include: The edge detection structured light field two-dimensional intensity division matrix generation unit is used to generate the edge detection structured light field two-dimensional intensity division matrix, the edge detection structured light field two-dimensional intensity division matrix is ​​obtained by fusing the Sobel edge detection operator with the Hardcode light field two-dimensional intensity distribution matrix, and the edge detection structured light field two-dimensional intensity division matrix generation method includes: Generate a two-dimensional intensity distribution matrix of the Haddon light field with a size of N×N; The size of each group of Hardcode light field two-dimensional intensity distribution matrix is ​​enlarged by 3 times; The pixel blocks with a scale of 3×3 and values ​​of all +1 on each enlarged two-dimensional intensity distribution matrix of the Haddon light field are replaced with the Sobel edge detection operator to obtain the two-dimensional intensity distribution matrix of the edge detection structured light field; An illumination beam generating unit, comprising a light source and a light field modulator, the light field modulator being loaded with the edge detection structure light field two-dimensional intensity division matrix, and being used for modulating the light beam generated by the light source, and irradiating the modulated light beam onto the target; A target echo acquisition unit, used to obtain the echo intensity corresponding to the target when the edge detection structured light field illuminates the target; The target edge contour extraction unit is used to reconstruct the target image based on the echo intensity, obtain the target gradient map based on the target image, and perform gradient detection based on the target gradient map to extract the target edge contour.

10. A target recognition device in a single pixel system, characterized in that: include: The target edge detection device in a single pixel system as claimed in claim 9, used to extract the edge contour of the target; Target edge contour feature extraction unit, used to use Fourier descriptor Characterize the edge contour features of each target; Recognition unit, based on Fourier descriptor Identify and distinguish the edge contours of different targets, and then distinguish different targets.

11. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the target edge detection method under the single pixel system as claimed in claim 1 are implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the target edge detection method under the single pixel system as claimed in claim 1 are implemented.