Target detection method and target detection system

Abstract

According to the target detection method and the target detection system, an improved self-adaptive FAST algorithm is introduced after wavelet packet transformation filtering, a threshold value is associated with a center pixel and a neighborhood pixel on the basis of an original algorithm, rapid target extraction is carried out, and information such as target center frequency and orientation is obtained. According to the method, information such as target center frequency and orientation in the K-space tomography reconstruction image after holographic radio imaging can be accurately and quickly detected, hardware acceleration is realized on the FPGA, and an efficient solution is provided for the target detection problem in channel state information perception.

Description

Technical Field

[0001] The present invention relates to the field of signal processing, and in particular to a target detection method and a target detection system based on microwave photon holographic radio imaging technology. Background Art

[0002] In current mobile communication technology, how to enable the imaging system to detect targets quickly and accurately in the presence of a large amount of noise is a major challenge.

[0003] The statements herein merely provide background art related to the present invention and do not necessarily constitute prior art. Summary of the invention

[0004] The purpose of the present invention is to provide a target detection method and a target detection system, which can accurately and quickly detect the target center frequency, orientation and other information in the K-space tomography reconstructed image after holographic radio imaging, and realize hardware acceleration, so as to provide an efficient solution to the target detection problem in channel state information perception.

[0005] In order to achieve the above object, the present invention provides a target detection method, comprising the following steps:

[0006] Step S1, determining the wavelet packet basis function and the number of decomposition layers;

[0007] The mathematical expression of the wavelet packet basis function is:

[0008]

[0009] in, is the logarithmic energy entropy of the i-th projection coefficient of the image on the wavelet basis function, j is the number of decomposition layers, and k is the order of the j-th layer;

[0010] Step S2, two-dimensional wavelet packet decomposition;

[0011] After determining the wavelet packet basis function and the number of decomposition layers, the fast Mallat algorithm is used to perform two-dimensional wavelet packet decomposition on the image. L represents the low-frequency coefficient, H represents the high-frequency coefficient, and the decomposition expression of the Mallat algorithm is:

[0012]

[0013] in, is the original signal to be processed, and are the approximate coefficients and exact coefficients obtained by wavelet packet decomposition, h n is the low frequency filter coefficient, g n is the high frequency filter coefficient;

[0014] Step S3, determining a threshold calculation criterion;

[0015] After the wavelet packet is decomposed into different wavelet packet coefficients, the threshold calculation criterion is determined for each wavelet packet coefficient;

[0016] The adaptive threshold criterion expression is as follows:

[0017]

[0018] in, is the noise mean square error, mid is the median value of the wavelet coefficients, λ b is the square of the wavelet coefficient corresponding to the minimum risk loudness element;

[0019] The fixed threshold criterion calculation expression is:

[0020]

[0021] Among them, σ is the noise mean square error, N is the length of each wavelet coefficient;

[0022] Heuristic threshold criterion: Calculate the adaptive threshold th r and a fixed threshold th s Then, order:

[0023]

[0024] Its expression is:

[0025]

[0026] The maximum and minimum threshold criteria are obtained by fixing the threshold, and the expression is:

[0027]

[0028] Step S4, determining a threshold processing function;

[0029] The mathematical expression of the hard threshold processing function is:

[0030]

[0031] Among them, th is the set threshold, w j,k is the kth wavelet coefficient of the jth layer, is the wavelet coefficient after threshold processing;

[0032] The mathematical expression of the soft threshold processing function is:

[0033]

[0034] The mathematical expression of the soft-hard compromise threshold processing function is:

[0035]

[0036] Among them, the parameter α∈[0,1], when α=0, it is a hard threshold processing function, and when α=1, it is a soft threshold processing function;

[0037] Step S5, wavelet packet two-dimensional reconstruction;

[0038] Step S6, adaptive threshold FAST algorithm;

[0039] Step S6.1: Take the central pixel with gray value P as the Bresenham circle with radius R, and make the difference between the central pixel and the pixels on the circle to get the gray difference value N. i =PP i ;

[0040] Step S6.2, calculating the threshold T corresponding to the central pixel;

[0041] The calculation formula for setting the threshold T is:

[0042]

[0043] Among them, P max is the maximum gray value of the circumferential pixels, P min is the minimum gray value among the circumferential pixels;

[0044] Step S6.3: Compare grayscale difference N i and the size of the adaptive threshold T and calculate N i The number of pixels is greater than T. When the number of pixels is greater than m, the central pixel is judged to be a candidate target;

[0045] Step S6.4: Perform non-maximum suppression on the candidate targets.

[0046] The present invention also provides a target detection system for implementing the target detection method, the target detection system is implemented based on an FPGA chip, and the target detection system comprises: a data storage module, a wavelet packet threshold denoising module, a target detection extraction module, and a data output module;

[0047] The data to be processed is input by the data storage module, and most of the noise is filtered out by the wavelet packet threshold denoising module. The target detection and extraction module detects the target information, and the detected target information is output through the data output module to achieve the acceleration of the target detection algorithm based on wavelet packet threshold denoising.

[0048] The present invention introduces an improved adaptive FAST algorithm after wavelet packet transform filtering, associates the threshold with the central pixel and the neighborhood pixel on the basis of the original algorithm, and performs rapid target extraction to obtain information such as the target center frequency and orientation. The present invention can accurately and quickly detect information such as the target center frequency and orientation in the K-space tomography reconstructed image after holographic radio imaging, and implements hardware acceleration on FPGA, providing an efficient solution to the target detection problem in channel state information perception. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flow chart of a target detection method provided by the present invention.

[0050] Figure 2 It is a two-dimensional wavelet packet decomposition structure diagram.

[0051] Figure 3 This is a schematic diagram of Bresenham Circle.

[0052] Figure 4 It is a structural block diagram of a target detection system provided by the present invention.

[0053] Figure 5 It is a tomographic reconstruction image in the K-space rectangular coordinate system.

[0054] Figure 6 It is the target detection map of the adaptive FAST algorithm.

[0055] Figure 7 It is the target detection map of FAST algorithm.

[0056] Figure 8a It is the image after boundary extension transformation.

[0057] Figure 8b This is the waveform of the data input into the Vivado software.

[0058] Figure 9a It is the waveform diagram of wavelet packet reconstruction and target detection result of Vivado software.

[0059] Figure 9b It is the input image for experimental verification of the target detection algorithm.

[0060] Fig.9c is the target two-dimensional wave vector spatial distribution after data processing. DETAILED DESCRIPTION

[0061] The following is based on Figure 1 9 , a preferred embodiment of the present invention is specifically described.

[0062] In 6G mobile communication technology, correct perception of channel state information can greatly improve communication performance, realize space division multiplexing, and make beamforming more accurate. Intelligent holographic radio imaging technology can accurately perceive the electromagnetic space of communication users in complex environments, and obtain interference patterns with extremely high spatial resolution and extremely fine spectral resolution. K-space tomography reconstruction can obtain information such as the center frequency azimuth of the incident RF signal.

[0063] like Figure 1 As shown, the present invention provides a target detection method, comprising the following steps:

[0064] Step S1, determining the wavelet packet basis function and the number of decomposition layers;

[0065] The wavelet packet energy entropy reflects the distribution of image energy at each scale. The wavelet basis function and decomposition layer number that are most suitable for the image to be processed are determined by calculating the wavelet packet energy entropy. The smaller the entropy value, the better the decomposition effect of the wavelet packet.

[0066] The mathematical expression of the wavelet packet basis function is:

[0067]

[0068] in, is the logarithmic energy entropy of the i-th projection coefficient of the image on the wavelet basis function, j is the number of decomposition layers, and k is the order of the j-th layer;

[0069] Step S2, two-dimensional wavelet packet decomposition;

[0070] After determining the wavelet packet basis function and the number of decomposition layers, the fast Mallat algorithm is used to perform two-dimensional wavelet packet decomposition on the image. The decomposition diagram is shown in the figure. Figure 2 As shown, L represents the low-frequency coefficient and H represents the high-frequency coefficient. The Mallat algorithm can convert the wavelet packet transform into the form of convolution operation of the signal and the filter bank. The algorithm can realize the decomposition and reconstruction of the specific signal based on the coefficients of the filter bank.

[0071] Mallat algorithm decomposes the expression into:

[0072]

[0073] in, is the original signal to be processed, and are the approximate coefficients and exact coefficients obtained by wavelet packet decomposition, h n is the low frequency filter coefficient, g n is the high frequency filter coefficient;

[0074] Step S3, determining a threshold calculation criterion;

[0075] After the wavelet packet is decomposed into different wavelet packet coefficients, the threshold calculation criterion is determined for each wavelet packet coefficient;

[0076] The adaptive threshold criterion expression is as follows:

[0077]

[0078] in, is the noise mean square error, mid is the median value of the wavelet coefficients, λ b is the square of the wavelet coefficient corresponding to the minimum risk loudness element;

[0079] The fixed (Sqtwolog) threshold criterion calculation expression is:

[0080]

[0081] Among them, σ is the noise mean square error, N is the length of each wavelet coefficient;

[0082] The heuristic threshold criterion calculates the adaptive threshold th r and a fixed threshold th s Then, order:

[0083]

[0084] Its expression is:

[0085]

[0086] The Minimaxi threshold criterion is obtained by fixing the threshold, and the expression is:

[0087]

[0088] Step S4, determining a threshold processing function;

[0089] After determining the threshold, it is necessary to determine the threshold processing function to process the wavelet packet coefficients accordingly. The selection of different threshold processing functions will also result in different continuity and accuracy of the reconstructed image.

[0090] The mathematical expression of the hard threshold processing function is:

[0091]

[0092] Among them, th is the set threshold, w j,k is the kth wavelet coefficient of the jth layer, is the wavelet coefficient after threshold processing;

[0093] The mathematical expression of the soft threshold processing function is:

[0094]

[0095] The mathematical expression of the soft-hard compromise threshold processing function is:

[0096]

[0097] Among them, the parameter α∈[0,1], when α=0, it is a hard threshold processing function, and when α=1, it is a soft threshold processing function;

[0098] Step S5, wavelet packet two-dimensional reconstruction;

[0099] When wavelet packets are reconstructed, column vectors are reconstructed;

[0100] Step S6, adaptive threshold FAST algorithm;

[0101] The steps of the adaptive FAST detection algorithm are as follows:

[0102] Step S6.1: Figure 3 As shown in the figure, the central pixel with gray value P is used as a Bresenham circle with radius R. The gray difference N can be obtained by subtracting the central pixel from the pixels on the circle. i =PP i ;

[0103] Step S6.2, calculating the threshold T corresponding to the central pixel;

[0104] The calculation formula for setting the threshold T is:

[0105]

[0106] Among them, P max is the maximum gray value of the circumferential pixels, P min is the minimum gray value among the circumferential pixels;

[0107] Step S6.3: Compare grayscale difference N i and the size of the adaptive threshold T and calculate N i The number of pixels is greater than T. When the number of pixels is greater than m, the central pixel is judged to be a candidate target;

[0108] Step S6.4: Perform non-maximum suppression on the candidate targets.

[0109] like Figure 4As shown, the present invention also provides a target detection system, which is implemented based on an FPGA chip. The target detection system includes: a data storage module, a wavelet packet threshold denoising module, a target detection extraction module, and a data output module. First, the data to be processed is input by the data storage module, and most of the noise is filtered out by the wavelet packet threshold denoising module (the wavelet packet threshold denoising module implements steps S1 to S4), and the target detection extraction module detects target information (the target detection extraction module implements step S5). Finally, the detected target information is output through the data output module to achieve acceleration of the target detection algorithm based on the wavelet packet threshold denoising.

[0110] Example

[0111] In one embodiment of the present invention, a perception experiment is conducted based on a holographic radio imaging system to verify the effectiveness of the target detection algorithm based on wavelet packet analysis provided by the present invention.

[0112] In the target perception scenario set in the experiment, the angle range is ±30° and the angle resolution is 0.33°, corresponding to K in K space. x Range is The frequency range is 30GHz-50GHz, and the frequency resolution is 0.1GHz, corresponding to K in K space. y Range is In the experiment, the 10 radiation sources emitted by the antenna are set with parameter information of (-28°, 46GHz), (-25°, 40GHz), (-20°, 33GHz), (-13°, 36GHz), (-2°, 37GHz), (-9°, 45GHz), (5°, 32GHz), (12°, 42GHz), (19°, 47GHz), and (25°, 38GHz). The K-space tomographic reconstruction results of holographic radio imaging are shown in Figure 2. Figure 5 , using the K-space rectangular coordinate system, the image resolution is 200×200.

[0113] According to the image features of holographic radio K-space tomography reconstruction, the optimal wavelet basis is calculated as db4 through wavelet packet energy entropy, and the optimal decomposition layer is one layer. The image is decomposed by the two-dimensional Mallat decomposition algorithm, and the image is denoised by using the fixed threshold criterion and soft threshold processing with better denoising effect in the simulation. The denoised wavelet packet coefficients are reconstructed by the two-dimensional Mallat algorithm. Finally, the adaptive FAST algorithm is used to detect targets on the wavelet packet reconstructed image. The radius of the central pixel Bresenham ring is set to 3, the number of pixels m = 12, and the adaptive FAST algorithm target detection diagram is shown in the figure. Figure 6 shown. Figure 7This is the detection result of the FAST algorithm. A total of 9 targets were detected. Among the 9 detected targets, the 9th target was a side peak in the peak neighborhood. The corresponding target was not detected correctly, resulting in a false alarm. The azimuth and center frequency information of the remaining targets were consistent with the set information. The adaptive FAST algorithm detected 10 set target information: (-28.240°, 45.932GHz), (-25.079°, 40.012GHz), (-20.213°, 33.088GHz), (-12.688°, 36.034GHz), (-9.152°, 45.021GHz), (-2°, 36.990GHz), (4.728°, 32.004GHz), (11.919°, 41.973GHz), (19.218°, 46.943GHz), (25.206°, 38.053GHz). The center frequency and azimuth information are consistent with the set information, the maximum angle error is 0.312°, and the maximum center frequency error is 0.0879 GHz, both within the resolution.

[0114] The present invention designs a hardware target detection system according to the target detection algorithm, realizes functional simulation, verifies the correctness and feasibility of the system, and its acceleration performance. The clock cycle is set to 10ns, the input data bit width is 10 bits, and the 214×200 image after the boundary extension of the image to be processed is shown in Figure 8(a). The input data waveform is shown in Figure 8(b) with one-dimensional hexadecimal input. The final detection result waveform is shown in Figure 9(a). The detection output data is transmitted to the PC and restored to image form, as shown in Figure 9(b). Figure 9(c) is the hardware input image. It can be seen that the hardware target detection system is consistent with the Matlab simulation results, and the center frequency and azimuth information of 10 targets are correctly detected, indicating the correctness of the functional simulation of the hardware platform and the hardware system design.

[0115] The wavelet packet threshold denoising module and the target detection extraction module in the target detection hardware system based on wavelet packet analysis both adopt a pipeline structure with row and column alignment. There is a total of 23 row cycle delays, or 4922 clock cycles. A total of 42,800 clock cycles are required for data input. The total operating time of the system is 0.24ms, or 47,722 clock cycles, which is about 230 times faster than the software platform, achieving acceleration of the target detection algorithm based on wavelet packet analysis.

[0116] The present invention introduces an improved adaptive FAST algorithm after wavelet packet transform filtering, associates the threshold with the center pixel and the neighborhood pixel on the basis of the original algorithm, and performs rapid target extraction to obtain information such as the center frequency and orientation of the target. The present invention implements FAST adaptive target detection based on the wavelet packet threshold denoising algorithm, associates the threshold size with the center pixel and its Bresenham ring pixel grayscale value, and improves the target detection accuracy by 20% relative to the FAST algorithm. The present invention uses a target detection hardware system to implement a target detection method based on wavelet packet transform, which is about dozens of times higher than the FAST detection algorithm, and can accelerate the target detection algorithm based on wavelet packet analysis in only milliseconds.

[0117] The present invention can accurately and quickly detect the target center frequency, orientation and other information in the K-space tomography reconstructed image after holographic radio imaging, and realize hardware acceleration on FPGA, providing an efficient solution to the target detection problem in channel state information perception.

[0118] It should be noted that, in the embodiments of the present invention, the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the embodiments, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0119] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0120] It should be understood that when used in this specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0121] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0122] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0123] As used in this specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0124] Although the content of the present invention has been described in detail through the above preferred embodiments, it should be appreciated that the above description should not be considered as limiting the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present invention. After reading the above content, it will be obvious for those skilled in the art to make various modifications and substitutions to the present invention. Therefore, the scope of protection of the present invention should be defined by the attached claims.

Claims

1. A target detection method, characterized in that: The following steps are involved: Step S1, determining the wavelet packet basis function and the number of decomposition layers; The mathematical expression of the wavelet packet basis function is: in, is the logarithmic energy entropy of the i-th projection coefficient of the image on the wavelet basis function, j is the number of decomposition layers, and k is the order of the j-th layer; Step S2, two-dimensional wavelet packet decomposition; After determining the wavelet packet basis function and the number of decomposition layers, the fast Mallat algorithm is used to perform two-dimensional wavelet packet decomposition on the image. L represents the low-frequency coefficient, H represents the high-frequency coefficient, and the decomposition expression of the Mallat algorithm is: in, is the original signal to be processed, and are the approximate coefficients and exact coefficients obtained by wavelet packet decomposition, h n is the low frequency filter coefficient, g n is the high frequency filter coefficient; Step S3, determining a threshold calculation criterion; After the wavelet packet is decomposed into different wavelet packet coefficients, the threshold calculation criterion is determined for each wavelet packet coefficient; The adaptive threshold criterion expression is as follows: in, is the noise mean square error, mid is the median value of the wavelet coefficients, λ b is the square of the wavelet coefficient corresponding to the minimum risk loudness element; The fixed threshold criterion calculation expression is: Among them, σ is the noise mean square error, N is the length of each wavelet coefficient; Heuristic threshold criterion: Calculate the adaptive threshold th r and a fixed threshold th s Then, order: Its expression is: The maximum and minimum threshold criteria are obtained by fixing the threshold, and the expression is: Step S4, determining a threshold processing function; The mathematical expression of the hard threshold processing function is: Among them, th is the set threshold, w j,k is the kth wavelet coefficient of the jth layer, is the wavelet coefficient after threshold processing; The mathematical expression of the soft threshold processing function is: The mathematical expression of the soft-hard compromise threshold processing function is: Among them, the parameter α∈[0,1], when α=0, it is a hard threshold processing function, and when α=1, it is a soft threshold processing function; Step S5, wavelet packet two-dimensional reconstruction; Step S6, adaptive threshold FAST algorithm; Step S6.1: Take the central pixel with gray value P as the Bresenham circle with radius R, and make the difference between the central pixel and the pixels on the circle to get the gray difference value N. i =PP i ; Step S6.2, calculating the threshold T corresponding to the central pixel; The calculation formula for setting the threshold T is: Among them, P max is the maximum gray value of the circumferential pixels, P min is the minimum gray value among the circumferential pixels; Step S6.3: Compare grayscale difference N i and the size of the adaptive threshold T and calculate N i The number of pixels is greater than T. When the number of pixels is greater than m, the central pixel is judged to be a candidate target; Step S6.4: Perform non-maximum suppression on the candidate targets.

2. A target detection system for implementing the target detection method according to claim 1, characterized in that: The target detection system is implemented based on an FPGA chip, and comprises: a data storage module, a wavelet packet threshold denoising module, a target detection extraction module, and a data output module; The data to be processed is input by the data storage module, and most of the noise is filtered out by the wavelet packet threshold denoising module. The target detection and extraction module detects the target information, and the detected target information is output through the data output module to achieve the acceleration of the target detection algorithm based on wavelet packet threshold denoising.

Images