Millimeter-wave image target detection method and system based on the distribution of hidden object positions

Abstract

The present application discloses a millimeter-wave image target detection method and system based on the location distribution of concealed objects. The millimeter-wave image target detection method based on the location distribution of concealed objects includes the following steps: statistically analyzing the distribution area of the central positions of concealed objects; determining the high-probability distribution area of concealed objects; slicing the high-probability distribution area; enhancing the super-resolution of the slices of the high-probability distribution area through a discrete wavelet transform super-resolution module; performing target detection on the original millimeter-wave image to obtain a first concealed object detection result; performing target detection on the slices of the high-probability distribution area to obtain a second concealed object detection result; and combining the first concealed object detection result and the second concealed object detection result. It has advantages such as effectively alleviating the contradiction between the accuracy and real-time performance of millimeter-wave body security inspection.

Description

Technical Field

[0001] This application relates to the field of millimeter-wave image detection, and particularly to a millimeter-wave image target detection method and system based on the position distribution of concealed objects. Background Art

[0002] Millimeter-wave holographic imaging technology is an advanced technology in the field of security at present. Without directly contacting the human body, it can effectively detect items hidden in various parts of the human body under clothing, especially non-metallic items. In addition, millimeter-wave human imaging devices have the characteristics of being harmless to the human body and having strong penetration. Although its transmission power is less than one-thousandth of the electromagnetic wave radiation of a mobile phone, it can effectively improve the objectivity, accuracy, and pertinence of inspections, reduce the labor intensity of security inspectors, and improve the security inspection efficiency.

[0003] Compared with visible light, the imaging resolution of millimeter-wave imaging technology is relatively low, and it is easily affected by the surrounding environment, forming clutter, which reduces the imaging quality. At the same time, concealed objects are usually small in size and have limited texture information of their own, making detection difficult. Existing millimeter-wave target detection technologies have a low detection rate for small targets, there is a risk of missed detection, and there are potential safety hazards.

[0004] At the same time, due to the particularity of security inspection tasks, it is necessary to output the detection results on the spot, and there is a high requirement for real-time performance. These limit the use of complex models with a large number of parameters. Any solution to improve the model detection performance must take into account its real-time performance requirements. Summary of the Invention

[0005] One aspect of the embodiments of this application provides a millimeter-wave image target detection method based on the position distribution of concealed objects, which may include the following steps:

[0006] Statistically analyze the distribution area of the central positions of concealed objects;

[0007] Determine the high-probability distribution area of concealed objects;

[0008] Slice the high-probability distribution area;

[0009] Through a discrete wavelet transform super-resolution module, perform super-resolution enhancement on the slices of the high-probability distribution area;

[0010] Perform target detection on the original millimeter-wave image to obtain the first concealed object detection result;

[0011] Perform target detection on the slices of the high-probability distribution area to obtain the second concealed object detection result;

[0012] Merge the first concealed object detection result and the second concealed object detection result.

[0013] The millimeter-wave image target detection method based on the location distribution of concealed objects in this application statistically analyzes the location distribution of concealed objects. According to the extremely uneven characteristics of the location distribution of concealed objects in millimeter-wave security inspection tasks, different detection means are adopted for different distribution situations; in the detection task of millimeter-wave image targets, the location distribution of concealed objects is not uniform, and most concealed objects are concentrated in specific areas. Therefore, the areas where these concealed objects are concentrated can be sliced, the slices can be enhanced with high resolution, and then the target detector is used to perform target detection on the original picture and the slices respectively, and finally the detection results of both are fused.

[0014] Slicing and enhancing the high-probability areas where concealed objects appear improves the detection ability of the model; at the same time, only high-resolution enhancement is performed on key areas, saving computational resources and reducing the model processing time.

[0015] Preferably, the step of performing super-resolution enhancement on the slices of the high-probability distribution area through the discrete wavelet transform super-resolution module includes the following steps:

[0016] Perform wavelet transform on the input image;

[0017] Stitch the wavelet components as a new feature vector;

[0018] Take the wavelet transform components of the low-resolution image as input and feed them into the convolutional neural network model to obtain the output convolutional neural network prediction difference;

[0019] Obtain the enhanced wavelet components;

[0020] Obtain the high-resolution image through inverse wavelet transform.

[0021] Through the above technical solution, the discrete wavelet transform super-resolution module (DWSR module) is divided into the following links: First, perform two-dimensional discrete wavelet transform on the input image to obtain wavelet components, and then take the wavelet components as a whole and feed them into the neural network model. The output of the neural network is the difference between the high-resolution image and the low-resolution image; add the difference between the high-resolution image and the low-resolution image to the sub-image of the original input image to obtain the enhanced sub-image; finally, perform inverse wavelet transform on the enhanced wavelet components to obtain the super-resolution enhanced slices.

[0022] Preferably, the step of performing wavelet transform on the input image includes the following steps:

[0023] Perform two-dimensional discrete wavelet transform on the input original image to obtain the wavelet components of low-frequency information, horizontal high-frequency information, vertical high-frequency information, and diagonal high-frequency information;

[0024] Through the two-dimensional discrete wavelet transform, the original image is transformed into wavelet components of low-frequency information, horizontal high-frequency information, vertical high-frequency information, and diagonal high-frequency information. The original image and the wavelet components are mutually converted through forward and inverse transforms;

[0025] The transformation method of the two-dimensional discrete wavelet includes formula (1):

[0026] , (1)

[0027] The inverse transformation method of the two-dimensional discrete wavelet includes formula (2):

[0028] , (2)

[0029] In formulas (1) and (2), A represents the first adjacent pixel value in the original image, B represents the second adjacent pixel value in the original image, C represents the third adjacent pixel value in the original image, D represents the fourth adjacent pixel value in the original image respectively, a represents the pixel value at the corresponding position in the wavelet component of low-frequency information, b represents the pixel value at the corresponding position in the wavelet component of horizontal high-frequency information, c represents the pixel value at the corresponding position in the wavelet component of vertical high-frequency information, and d represents the pixel value at the corresponding position in the wavelet component of diagonal high-frequency information.

[0030] Preferably, the splicing method of splicing the wavelet components as a new feature vector includes:

[0031] Splice the wavelet components of low-frequency information, horizontal high-frequency information, vertical high-frequency information, and diagonal high-frequency information in the channel dimension to obtain a new feature vector. The obtaining method of the new feature vector is as shown in formula (3):

[0032] , (3)

[0033] In the formula, represents the new feature vector, LL represents the wavelet component of low-frequency information, LH represents the horizontal high-frequency information, HL represents the wavelet component of vertical high-frequency information, HH represents the wavelet component of diagonal high-frequency information, represents the two-dimensional discrete wavelet transform of the low-resolution original image.

[0034] Preferably, the step of taking the wavelet transform components of the low-resolution image as inputs and feeding them into a convolutional neural network model to obtain the output convolutional neural network prediction difference includes the following steps:

[0035] Adopt a discrete wavelet transform super-resolution neural network model (i.e., DWSR neural network model), and use the difference between the high-resolution image and the low-resolution image as the true value for training the discrete wavelet transform super-resolution neural network model;

[0036] Use the wavelet transform components of the low-resolution image as the input, and use a convolutional neural network to predict the difference between the high-resolution image and the low-resolution image. The difference between the high-resolution image and the low-resolution image is as shown in formula (4):

[0037] , (4)

[0038] In the formula, represents the difference between the high-resolution image and the low-resolution image, respectively represent the difference in the low-frequency dimension, the difference in the horizontal high-frequency dimension, the difference in the vertical high-frequency dimension, and the difference in the diagonal high-frequency dimension.

[0039] Preferably, the method for obtaining the enhanced wavelet components includes:

[0040] Add and sum the new feature vector and the difference between the high-resolution image and the low-resolution image output by the convolutional neural network to obtain the enhanced wavelet components:

[0041] , (5)

[0042] In the formula, represents the new feature vector, represents the difference between the high-resolution image and the low-resolution image, respectively represent the difference in the low-frequency dimension, the difference in the horizontal high-frequency dimension, the difference in the vertical high-frequency dimension, and the difference in the diagonal high-frequency dimension, LL represents the low-frequency information wavelet component, LH represents the horizontal high-frequency information, HL represents the vertical high-frequency information wavelet component, HH represents the diagonal high-frequency information wavelet component.

[0043] Preferably, the method for obtaining the high-resolution image through inverse wavelet transform includes formula (6):

[0044] , (6)

[0045] In the formula, HR represents the high-resolution image, 2 dIDWT {.} represents the two-dimensional discrete inverse wavelet transform function, represents the new feature vector, represents the difference between the high-resolution image and the low-resolution image.

[0046] Preferably, the step of combining the first concealed object detection result and the second concealed object detection result includes the following steps:

[0047] Restore the position coordinates of the high-probability distribution area slice detection result to the coordinate system of the original millimeter-wave image;

[0048] Adopt the non-maximum suppression algorithm NMS to combine the first concealed object detection result and the second concealed object detection result.

[0049] Preferably, the step of counting the distribution area of the center position of the concealed object includes the following steps:

[0050] Divide the concealed object distribution map into several grids of preset pixels;

[0051] Traverse the data set and count the number of center points of the concealed objects in the labeled samples that appear in each grid.

[0052] Another aspect of the embodiments of the present application provides a millimeter-wave image target detection system based on the position distribution of concealed objects, which may include:

[0053] A statistical module for counting the distribution area of the center position of the concealed object;

[0054] A determination module for determining the high-probability distribution area of the concealed object;

[0055] A slicing module for slicing the high-probability distribution area;

[0056] A super-resolution enhancement module for enhancing the super-resolution of the high-probability distribution area slice through a discrete wavelet transform super-resolution module;

[0057] A first concealed object detection module for performing target detection on the original millimeter-wave image to obtain a first concealed object detection result;

[0058] A second concealed object detection module for performing target detection on the high-probability distribution area slice to obtain a second concealed object detection result;

[0059] A detection result merging module for merging the first concealed object detection result and the second concealed object detection result.

[0060] The embodiments of the present application have at least the following beneficial effects:

[0061] 1. In the detection task of millimeter wave image targets, the position distribution of hidden objects is not uniform, and most hidden objects appear in specific areas. Therefore, these areas can be sliced ​​and the slices can be enhanced with high resolution. Then, the target detector can be used to perform target detection on the original image and the slices respectively, and finally the detection results of the original image and the slices are fused. The present application is based on a millimeter wave image target detection method based on the position distribution of hidden objects. The high probability area where hidden objects appear is sliced ​​and enhanced, which improves the detection of hidden objects, especially the high detection rate of small targets. At the same time, only the key areas are enhanced with high resolution, which saves the amount of calculation and compresses the processing time of the model.

[0062] 2. This application is based on the millimeter wave image target detection method of the position distribution of hidden objects. According to the characteristics of the huge difference in the distribution probability of hidden objects in different positions in the millimeter wave security inspection task, the computing resources are allocated in a targeted manner. By statistically analyzing the position distribution of the center point of the hidden object, the high probability area of ​​the hidden object is determined, and super-resolution enhancement and detection are carried out in a targeted manner. This application not only effectively improves the detection rate of millimeter wave human body security inspection, but also reduces the overall calculation amount, saves time, improves real-time performance, and effectively alleviates the contradiction between the accuracy and real-time performance of millimeter wave human body security inspection. It has great practical significance for the widespread promotion and application of millimeter wave security inspection technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flow chart of the millimeter wave image target detection method based on the location distribution of hidden objects in this application.

[0064] Figure 2 This is a schematic diagram of the principle of the millimeter wave image target detection method based on the location distribution of hidden objects in this application.

[0065] Figure 3 This is a distribution statistical map of the central location of hidden objects in this application.

[0066] Figure 4a This is a schematic diagram of a specific location where most objects in this application are concentrated.

[0067] Figure 4b Schematic diagram of the four areas of high probability distribution of objects defined for this application.

[0068] Figure 5 This is a slice map of the high probability distribution area for this application.

[0069] Figure 6 This is a schematic diagram of the DWSR module structure of this application.

[0070] Figure 7 This is a schematic diagram of the mutual conversion between discrete wavelet transform and inverse discrete wavelet transform in this application. DETAILED DESCRIPTION

[0071] Embodiment

[0072] The various solutions and features of the present application are described herein with reference to the accompanying drawings.

[0073] The millimeter-wave image target detection method based on the distribution of hiding object positions in this embodiment, as Figure 1 shown, may specifically include the following steps:

[0074] S1. Statistically analyze the distribution area of the central positions of the hiding objects;

[0075] S2. Determine the high-probability distribution area of the hiding objects;

[0076] S3. Slice the high-probability distribution area;

[0077] S4. Through the discrete wavelet transform super-resolution module, perform super-resolution enhancement on the slices of the high-probability distribution area;

[0078] S5. Perform target detection on the original millimeter-wave image to obtain the first hiding object detection result;

[0079] S6. Perform target detection on the slices of the high-probability distribution area to obtain the second hiding object detection result;

[0080] S7. Combine the first hiding object detection result and the second hiding object detection result.

[0081] In the millimeter-wave image target detection task, the distribution of hiding object positions is not uniform, and most hiding objects are concentrated in specific areas. The millimeter-wave image target detection method based on the distribution of hiding object positions in this embodiment slices these specific areas where most hiding objects are concentrated, performs high-resolution enhancement on the slices, and then uses the target detector to perform target detection on the original image and the slices respectively. Finally, the detection results of the original image and the slices are fused, as Figure 2 shown.

[0082] The specific solution is as follows:

[0083] Step S1: Statistically analyze the distribution of the central positions of the hiding objects.

[0084] As Figure 3 shown, divide the distribution map into 10×10 pixel grids, traverse the data set, record the number of hiding object center points that appear in each grid in the labeled samples, and then visualize the distribution map. The higher the number of hiding objects corresponding to each grid in the distribution map, the higher the brightness.

[0085] Step S2: Determine the high-probability distribution area

[0086] Analyze the statistical chart of the distribution of hidden objects. Most of the hidden objects are concentrated in several specific positions, namely, both sides of the chest and abdomen and both sides of the thighs. As Figure 4a shown, the areas where most of the hidden objects are concentrated are delimited into four areas, as Figure 4b shown by the four squares in

[0087] These are called high-probability distribution areas. After statistics, more than 70% of the hidden objects fall in this high-probability distribution area.

[0088] Step S3: Slice the high-probability area Figure 5 According to the position of the high-probability area, slice the original input image to obtain four local images, as

[0089] shown.

[0090] Step S4: Use the Discrete Wavelet transform Super-Resolution module (abbreviated as DWSR) to enhance the super-resolution of the slices of the high-probability area. Figure 6 shown.

[0091] Step S41: Perform wavelet transform on the input image

[0092] Perform the Haar two-dimensional discrete wavelet transform (i.e., 2dDWT) on the input image LR to obtain four wavelet components: the low-frequency information wavelet component LL, the horizontal high-frequency information wavelet component LH, the vertical high-frequency information wavelet component HL, and the diagonal high-frequency information wavelet component HH.

[0093] The Haar two-dimensional discrete wavelet transform is as Figure 7 shown. Figure 7 The left side of Figure 7On the right side are the four wavelet components of the transformed low-frequency information wavelet component LL, horizontal high-frequency information wavelet component LH, vertical high-frequency information wavelet component HL, and diagonal high-frequency information wavelet component HH; the input original image and components can be mutually converted through transformation and inverse transformation; where A, B, C, and D represent four adjacent pixel values in the input original image, and a, b, c, and d respectively represent the pixel values at the corresponding positions in the four wavelet components of the low-frequency information wavelet component LL, horizontal high-frequency information wavelet component LH, vertical high-frequency information wavelet component HL, and diagonal high-frequency information wavelet component HH.

[0094] The transformation method of the two-dimensional discrete wavelet includes formula (1):

[0095] , (1)

[0096] The inverse transformation method of the two-dimensional discrete wavelet includes formula (2):

[0097] , (2)

[0098] In formulas (1) and (2), A represents the first adjacent pixel value in the original image, B represents the second adjacent pixel value in the original image, C represents the third adjacent pixel value in the original image, D respectively represents the fourth adjacent pixel value in the original image, a represents the pixel value at the corresponding position in the wavelet component of the low-frequency information, b represents the pixel value at the corresponding position in the wavelet component of the horizontal high-frequency information, c represents the pixel value at the corresponding position in the wavelet component of the vertical high-frequency information, and d represents the pixel value at the corresponding position in the wavelet component of the diagonal high-frequency information.

[0099] Step S42: Concatenate the wavelet components as a new feature vector

[0100] Concatenate the four wavelet components of the low-frequency information wavelet component LL, horizontal high-frequency information wavelet component LH, vertical high-frequency information wavelet component HL, and diagonal high-frequency information wavelet component HH in the channel dimension to obtain a new feature vector (Low Resolution Sub Bands, abbreviated as: LRSB), the new feature vector The obtaining method of is as formula (3):

[0101] , (3)

[0102] In the formula, represents the new feature vector, LL represents the low-frequency information wavelet component, LH represents the horizontal high-frequency information, HL represents the vertical high-frequency information wavelet component, HH represents the diagonal high-frequency information wavelet component, Represents the two-dimensional discrete wavelet transform of a low-resolution original image.

[0103] Step S43: Feed the LRSB as input into a convolutional neural network model, and the convolutional neural network model outputs the difference between the high-resolution image and the low-resolution image .

[0104] The DWSR neural network model takes the difference between the high-resolution image and the low-resolution image as the ground truth for model training, takes the wavelet transform component LRSB of the low-resolution image LR as input, and uses a convolutional neural network to predict the difference between the high-resolution image and the low-resolution image .

[0105] The DWSR neural network model takes the difference between the high-resolution image and the low-resolution image as the ground truth for model training, takes the wavelet transform component LRSB of the low-resolution image LR as input, and uses a convolutional neural network to predict the difference between the high-resolution image and the low-resolution image , and the difference between the high-resolution image and the low-resolution image predicted by the convolutional neural network is as shown in formula (4):

[0106] , (4)

[0107] In the formula, represents the difference between the high-resolution image and the low-resolution image, respectively represent the differences in the low-frequency dimension, horizontal high-frequency dimension, vertical high-frequency dimension, and diagonal high-frequency dimension.

[0108] This embodiment uses a pre-trained DWSR convolutional neural network model.

[0109] Step S44: Obtain the enhanced wavelet component

[0110] Add the new feature vector LRSB and the difference between the high-resolution image and the low-resolution image output by the convolutional neural network to sum, and obtain the enhanced wavelet component:

[0111] , (5)

[0112] In the formula, represents the new feature vector, represents the difference between the high-resolution image and the low-resolution image, respectively represent the differences in the low-frequency dimension, horizontal high-frequency dimension, vertical high-frequency dimension, and diagonal high-frequency dimension, LLRepresents the wavelet component of low-frequency information, LH Represents the horizontal high-frequency information, HL Represents the wavelet component of vertical high-frequency information, HH Represents the wavelet component of diagonal high-frequency information.

[0113] Step S45: Obtain the high-resolution image through the inverse transform method of two-dimensional discrete wavelet

[0114] The inverse transform method of two-dimensional discrete wavelet is shown in Formula (2). The method for obtaining the high-resolution image HR through the inverse transform of two-dimensional discrete wavelet includes Formula (6):

[0115] , (6)

[0116] In the formula, HR Represents the high-resolution image, 2 dIDWT {.} represents the two-dimensional discrete wavelet inverse transform function, Represents the new feature vector, Represents the difference between the high-resolution image and the low-resolution image.

[0117] Step S5: Perform target detection on the original millimeter-wave image to obtain the first hidden object detection result.

[0118] Step S6: Perform target detection on the sliced high-probability distribution region to obtain the second hidden object detection result;

[0119] Step S7: Combine the first hidden object detection result and the second hidden object detection result.

[0120] The step of combining the first hidden object detection result and the second hidden object detection result includes the following steps:

[0121] Restore the position coordinates of the sliced high-probability distribution region detection result to the coordinate system of the original millimeter-wave image;

[0122] Adopt the non-maximum suppression algorithm NMS to combine the first hidden object detection result and the second hidden object detection result.

[0123] It should be understood that various modifications can be made to the embodiments applied herein. Therefore, the above specification should not be regarded as a limitation, but only as an example of the embodiments. Those skilled in the art will think of other modifications within the scope and spirit of the present application.

[0124] The drawings included in the specification and constituting a part of the specification illustrate the embodiments of the present application, and together with the general description of the present application given above and the detailed description of the embodiments given below are used to explain the principles of the present application.

[0125] These and other features of the present application will become apparent from the following description of the preferred forms of the embodiments given as non - limiting examples with reference to the accompanying drawings.

[0126] It should also be understood that although the present application has been described with reference to some specific examples, those skilled in the art can surely implement many other equivalent forms of the present application.

[0127] When combined with the accompanying drawings, the above - mentioned and other aspects, features and advantages of the present application will become more apparent in view of the following detailed description.

[0128] Specific embodiments of the present application will be described hereinafter with reference to the accompanying drawings; however, it should be understood that the claimed embodiments are merely examples of the present application and can be implemented in various ways. Well - known and / or repetitive functions and structures are not described in detail to avoid obscuring the present application with unnecessary or redundant details. Therefore, the specific structural and functional details claimed herein are not intended to be limiting, but rather serve as a basis for the claims and a representative basis for teaching those skilled in the art to use the present application in substantially any suitable detailed structure in a variety of ways.

[0129] This specification may use the phrases "in one embodiment", "in another embodiment", "in yet another embodiment" or "in other embodiments", each of which may refer to one or more of the same or different embodiments according to the present application.

[0130] The above embodiments are only exemplary embodiments of the present application and are not used to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art can make various modifications or equivalent replacements within the spirit and protection scope of the present application, and such modifications or equivalent replacements should also be regarded as falling within the protection scope of the present application.

Claims

1. A millimeter wave image target detection method based on hidden object position distribution, characterized in that: The following steps are involved: Counting the distribution area of ​​the center position of the hidden object includes: dividing the hidden object distribution map into a number of grids of preset pixels; traversing the data set, counting the number of the center points of the hidden objects in the labeled samples appearing in each grid, and visualizing the distribution map. The more the number of hidden objects corresponding to each grid in the distribution map, the higher the brightness; Determining the high probability distribution area of ​​hidden objects, including: analyzing the hidden object distribution statistical map, and defining the area where most hidden objects are concentrated as the high probability distribution area of ​​hidden objects; Slicing the high probability distribution area; Through a discrete wavelet transform super-resolution module, super-resolution enhancement is performed only on the slices in the high probability distribution area; Performing target detection on the original millimeter wave image to obtain a first hidden object detection result; Performing target detection on the high probability distribution area slice to obtain a second hidden object detection result; The first hidden object detection result and the second hidden object detection result are combined to fuse the detection results of the original image and the slice.

2. The method according to claim 1, characterized in that The step of performing super-resolution enhancement on the high probability distribution area slice by using a discrete wavelet transform super-resolution module comprises the following steps: The input image is subjected to wavelet transform; The wavelet components are concatenated as new feature vectors; The wavelet transform component of the low-resolution image is fed into the convolutional neural network model as input to obtain the output convolutional neural network prediction difference; Get the enhanced wavelet component; The high-resolution image is obtained by inverse wavelet transform.

3. The method according to claim 2, characterized in that The step of performing wavelet transform on the input image comprises the following steps: Perform a two-dimensional discrete wavelet transform on the input original image to obtain a wavelet component of low-frequency information, a wavelet component of horizontal high-frequency information, a wavelet component of vertical high-frequency information, and a wavelet component of diagonal high-frequency information; By means of the two-dimensional discrete wavelet transform, the original image is transformed into a wavelet component of low-frequency information, a wavelet component of horizontal high-frequency information, a wavelet component of vertical high-frequency information and a wavelet component of diagonal high-frequency information, and the original image and the wavelet components are mutually converted by means of transform and inverse transform; The two-dimensional discrete wavelet transformation method includes formula (1): ，（1） The inverse transform method of the two-dimensional discrete wavelet includes formula (2): ， （2） In formula (1) and formula (2), A represents the first adjacent pixel value in the original image, B represents the second adjacent pixel value in the original image, C represents the third adjacent pixel value in the original image, and D represents the fourth adjacent pixel value in the original image. a represents the pixel value at the corresponding position in the wavelet component of the low-frequency information, b represents the pixel value at the corresponding position in the wavelet component of the horizontal high-frequency information, c represents the pixel value at the corresponding position in the wavelet component of the vertical high-frequency information, and d represents the pixel value at the corresponding position in the wavelet component of the diagonal high-frequency information.

4. The method according to claim 2, characterized in that: The method for splicing the wavelet components as a new feature vector includes: The low-frequency information wavelet component, the horizontal high-frequency information wavelet component, the vertical high-frequency information wavelet component and the diagonal high-frequency information wavelet component are concatenated in the channel dimension to obtain a new feature vector. The method for obtaining the new feature vector is as shown in formula (3): ，（3） In the formula, represents the new feature vector, LL represents the low-frequency information wavelet component, LH Represents horizontal high-frequency information, HL represents the vertical high-frequency information wavelet component, HH represents the diagonal high-frequency information wavelet component, Represents the 2D discrete wavelet transform of the low-resolution original image.

5. The method according to claim 2, characterized in that: The step of sending the wavelet transform component of the low-resolution image as input into the convolutional neural network model to obtain the output convolutional neural network prediction difference includes the following steps: Using a discrete wavelet transform super-resolution neural network model, the difference between the high-resolution image and the low-resolution image is used as the true value for training the discrete wavelet transform super-resolution neural network model; The wavelet transform component of the low-resolution image is used as input, and the convolutional neural network is used to predict the difference between the high-resolution image and the low-resolution image. The difference between the high-resolution image and the low-resolution image predicted by the convolutional neural network is as shown in formula (4): ，（4） In the formula, Represents the difference between the high-resolution image and the low-resolution image. They respectively represent the difference in the low-frequency dimension, the difference in the horizontal high-frequency dimension, the difference in the vertical high-frequency dimension, and the difference in the diagonal high-frequency dimension.

6. The method according to claim 2, characterized in that The method for obtaining the enhanced wavelet component comprises: The new feature vector is added to the difference between the high-resolution image and the low-resolution image output by the convolutional neural network to obtain the enhanced wavelet component: ，（5） In the formula, represents the new feature vector, Represents the difference between the high-resolution image and the low-resolution image. They represent the difference in low-frequency dimension, horizontal high-frequency dimension, vertical high-frequency dimension, and diagonal high-frequency dimension, respectively. LL represents the low-frequency information wavelet component, LH Represents horizontal high-frequency information, HL represents the vertical high-frequency information wavelet component, HH Represents the diagonal high-frequency information wavelet component.

7. The method according to claim 2, characterized in that The method for obtaining a high-resolution image by inverse wavelet transform includes formula (6): ，（6） In the formula, HR represents a high-resolution image, 2 DWT {.} represents the two-dimensional discrete wavelet inverse transform function, represents the new feature vector, Represents the difference between a high-resolution image and a low-resolution image.

8. The method according to claim 1, characterized in that The step of merging the first hidden object detection result and the second hidden object detection result comprises the following steps: Restoring the position coordinates of the high probability distribution area slice detection result to the coordinate system of the original millimeter wave image; The first hidden object detection result and the second hidden object detection result are combined by using a non-maximum suppression algorithm.

9. A millimeter wave image target detection system based on hidden object location distribution, characterized in that: include: A statistical module, used for calculating the distribution area of ​​the center position of the hidden object according to the method of claim 1; A determination module, used to determine the high probability distribution area of ​​hidden objects; A slicing module, used for slicing the high probability distribution area; A super-resolution enhancement module, used for performing super-resolution enhancement on the high probability distribution area slices through a discrete wavelet transform super-resolution module; A first hidden object detection module, used to perform target detection on the original millimeter wave image to obtain a first hidden object detection result; A second hidden object detection module, used to perform target detection on the high probability distribution area slice to obtain a second hidden object detection result; The detection result merging module is used to merge the first hidden object detection result and the second hidden object detection result.

Images