Terahertz near-field radar image enhancement method

Abstract

The invention discloses a terahertz near-field radar image enhancement method, which comprises the following steps of: performing data conversion and normalization on a terahertz near-field radar image to be enhanced to obtain a normalized terahertz near-field radar image; performing stationary wavelet transform decomposition on the normalized terahertz near-field radar image to obtain a plurality of low-frequency sub-bands and a plurality of high-frequency sub-bands; performing high-frequency multi-direction filtering on the high-frequency sub-bands by using filters in different directions to obtain a plurality of enhanced high-frequency sub-bands; performing low-frequency adaptive linear enhancement on the low-frequency sub-bands by using the mean value and the standard deviation of the low-frequency sub-bands to obtain a plurality of enhanced low-frequency sub-bands; performing inverse stationary wavelet transform on the plurality of enhanced high-frequency sub-bands and the plurality of enhanced low-frequency sub-bands to obtain a reconstructed terahertz near-field radar image; and carrying out linear transformation on the reconstructed terahertz near-field radar image to obtain an enhanced terahertz near-field radar image. The method can enhance contrast, highlight edge features and enrich detail information.

Description

Technical Field

[0001] This invention relates to the field of image enhancement technology, and more specifically to a method for enhancing terahertz near-field radar images. Background Technology

[0002] Currently, terahertz waves, with their outstanding material penetration and anti-interference characteristics in the electromagnetic spectrum, as well as their relatively unaffected propagation by atmospheric conditions, have demonstrated significant application value in target detection. Compared to traditional microwave radar systems, terahertz radar systems possess wider operating bandwidth, higher spatial resolution, and stronger Doppler effect sensitivity, making them a cutting-edge technology for modern target detection, trajectory tracking, and feature recognition. With the continuous iteration of terahertz imaging algorithms, relatively clear imaging results can now be achieved, providing a fundamental support for subsequent image processing.

[0003] However, due to limitations in the intensity of terahertz light sources and differences in the reflectivity of target materials, existing imaging results generally suffer from low contrast, blurred edge features, and weak detail information. Furthermore, because the target and background exhibit similar brightness distributions in terahertz images, feature information is easily masked when the background brightness is close to that of the target, severely impacting the accuracy of subsequent target detection and feature recognition. Current image enhancement algorithms are mainly divided into two categories: deep learning-based methods (such as super-resolution reconstruction and generative adversarial networks) can effectively reconstruct detailed features, but are hampered by the high cost of terahertz imaging systems leading to a scarcity of training samples, resulting in difficulties in model convergence and artifacts; traditional methods (such as histogram equalization, Gaussian filtering, mean filtering, and bilateral filtering) can improve contrast or suppress noise, but can cause excessive noise enhancement, overexposure / underexposure in local areas, blurred edges, and loss of high-frequency details, failing to meet the high-quality enhancement requirements of terahertz near-field radar images.

[0004] Therefore, how to provide a method for enhancing terahertz near-field radar images that can enhance the contrast of the terahertz near-field radar image to be enhanced, highlight the edge features of the terahertz near-field radar image to be enhanced, and enrich the detailed information of the terahertz near-field radar image to be enhanced is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the object of the present invention is to provide a method for enhancing terahertz near-field radar images.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for enhancing terahertz near-field radar images includes the following steps:

[0008] S1: Perform data conversion and normalization on the enhanced terahertz near-field radar image to obtain a normalized terahertz near-field radar image;

[0009] S2: Perform stationary wavelet transform decomposition on the normalized terahertz near-field radar image to obtain several low-frequency sub-bands and several high-frequency sub-bands.

[0010] S3: High-frequency subbands are filtered in multiple directions using filters of different directions to obtain several enhanced high-frequency subbands;

[0011] The mean and standard deviation of the low-frequency subband are used to perform low-frequency adaptive linear enhancement of the low-frequency subband, resulting in several enhanced low-frequency subbands.

[0012] S4: Perform inverse stationary wavelet transform on the several enhanced high-frequency subbands and the several enhanced low-frequency subbands to obtain the reconstructed terahertz near-field radar image;

[0013] S5: Perform a linear transformation on the reconstructed terahertz near-field radar image to obtain an enhanced terahertz near-field radar image.

[0014] Preferably, the data conversion in S1 converts the pixel values ​​of all pixels in the terahertz near-field radar image to be enhanced from 8-bit unsigned integers to 32-bit floating-point numbers;

[0015] The normalization in S1 linearly maps the pixel values ​​of all pixels in the enhanced terahertz near-field radar image from the interval [0, 255] to the interval [0, 1].

[0016] Preferably, S2 specifically includes the following steps:

[0017] S21: The normalized terahertz near-field radar image is decomposed into a first-level decomposition using the low-pass and high-pass filters of the Haar filter bank to obtain the low-frequency subband LL1, the horizontal high-frequency subband LH1, the vertical high-frequency subband HL1, and the diagonal high-frequency subband HH1.

[0018] S22: The low-frequency subband LL1 is decomposed in the second stage using the low-pass and high-pass filters of the Haar filter bank to obtain the low-frequency subband LL2, the horizontal high-frequency subband LH2, the vertical high-frequency subband HL2, and the diagonal high-frequency subband HH2.

[0019] Preferably, the calculation formulas for the low-frequency subband LL1, the horizontal high-frequency subband LH1, the vertical high-frequency subband HL1, the diagonal high-frequency subband HH1, the low-frequency subband LL2, the horizontal high-frequency subband LH2, the vertical high-frequency subband HL2, and the diagonal high-frequency subband HH2 are as follows:

[0020] ;

[0021] ;

[0022] ;

[0023] ;

[0024] ;

[0025] ;

[0026] ;

[0027] ;

[0028] in, This represents the normalized terahertz near-field radar image; and The numbers represent the low-pass and high-pass filters of the Haar filter bank, respectively; * represents convolution operation; and T represents transpose.

[0029] Preferably, high-frequency subbands are subjected to multi-directional high-frequency filtering using filters in different directions to obtain several enhanced high-frequency subbands, based on the following formula:

[0030] ;

[0031] ;

[0032] ;

[0033] ;

[0034] ;

[0035] ;

[0036] Where * represents the convolution operation; K r Let K1 represent the r-th directional filter, where r = 1, 2, ..., 8. K1 represents the horizontal directional filter, K2 represents the vertical directional filter, K3 represents the 45° diagonal directional filter, K4 represents the 135° diagonal directional filter, K5 represents the horizontal edge enhancement filter, K6 represents the vertical edge enhancement filter, K7 represents the Laplace filter, and K8 represents the sharpening filter. This indicates that the enhanced horizontal high-frequency subband LH1 is obtained by performing high-frequency multi-directional filtering on the horizontal high-frequency subband LH1. This indicates the enhanced vertical high-frequency subband obtained by performing high-frequency multi-directional filtering on the vertical high-frequency subband HL1. This indicates that the enhanced diagonal high-frequency subband HH1 is obtained by performing high-frequency multi-directional filtering on the diagonal high-frequency subband HH1. This indicates that the enhanced horizontal high-frequency subband LH2 is obtained by performing high-frequency multi-directional filtering on the horizontal high-frequency subband LH2. This indicates the enhanced vertical high-frequency subband obtained by performing high-frequency multi-directional filtering on the vertical high-frequency subband HL2. This indicates that the enhanced diagonal high-frequency subband HH2 is obtained by performing high-frequency multi-directional filtering.

[0037] Preferably, the low-frequency sub-bands are subjected to low-frequency adaptive linear enhancement using the mean and standard deviation of the low-frequency sub-bands to obtain several enhanced low-frequency sub-bands, based on the following formula:

[0038] ;

[0039] ;

[0040] ;

[0041] ;

[0042] ;

[0043] ;

[0044] ;

[0045] ;

[0046] ;

[0047] ;

[0048] ;

[0049] ;

[0050] ;

[0051] ;

[0052] ;

[0053] ;

[0054] Where M represents the number of rows of the low-frequency subband LL1 and the low-frequency subband LL2; N represents the number of columns of the low-frequency subband LL1 and the low-frequency subband LL2; This represents the pixel value of the pixel in the i-th row and j-th column of the low-frequency subband LL1; and These represent the mean and standard deviation of the low-frequency subband LL1, respectively. , , These represent the mean adjustment coefficient, standard deviation adjustment coefficient, and global adjustment coefficient of the low-frequency subband LL1, respectively. This represents the adaptive enhancement coefficient of the low-frequency subband LL1; This represents the default enhancement factor of the low-frequency subband LL1; This represents the total enhancement factor of the low-frequency subband LL1; This indicates that the enhanced low-frequency subband LL1 is obtained by performing low-frequency adaptive linear enhancement. This represents the pixel value of the pixel in the i-th row and j-th column of the low-frequency subband LL2; and These represent the mean and standard deviation of the low-frequency subband LL2, respectively. , , These represent the mean adjustment coefficient, standard deviation adjustment coefficient, and global adjustment coefficient of the low-frequency subband LL2, respectively. This represents the adaptive enhancement coefficient of the low-frequency subband LL2; This represents the default enhancement factor of the low-frequency subband LL2; This represents the total enhancement factor of the low-frequency subband LL2; This indicates that the low-frequency subband LL2 is subjected to low-frequency adaptive linear enhancement, resulting in the enhanced low-frequency subband.

[0055] Preferably, S4 is implemented based on the following formula:

[0056] ;

[0057] in, The image represents the reconstructed terahertz near-field radar image; * indicates convolution operation.

[0058] Preferably, S5 is implemented based on the following formula:

[0059] ;

[0060] in, This refers to the enhanced terahertz near-field radar image.

[0061] Preferably, the above-mentioned method for enhancing terahertz near-field radar images further includes the following steps:

[0062] S6: The enhanced terahertz near-field radar image is locally contrast optimized using a limited contrast histogram equalization algorithm to obtain the final enhanced terahertz near-field radar image.

[0063] Preferably, S6 specifically includes the following steps;

[0064] S61: Divide the enhanced terahertz near-field radar image into m×n local regions; wherein, the size of each local region is... m and n are positive integers;

[0065] S62: Calculate the gray-level histogram of the s-th local region. Where k represents the gray level; ; ;

[0066] S63: Gray-level histogram Threshold cropping is performed to obtain the cropped grayscale histogram. ;in, Indicates taking The smaller value between and 2;

[0067] S64: Calculate the cumulative distribution function of the s-th local region. ;

[0068] S65: Set pixel value Substitute In k, obtain enhanced pixels ;in, express The pixel value of the e-th pixel; The e-th pixel is located within the s-th local region; Enhanced pixels This represents the pixel value of the e-th pixel in the s-th local region of the final enhanced terahertz near-field radar image.

[0069] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a method for enhancing terahertz near-field radar images, which can enhance the contrast of the terahertz near-field radar image to be enhanced, highlight the edge features of the terahertz near-field radar image to be enhanced, and enrich the detailed information of the terahertz near-field radar image to be enhanced. Attached Figure Description

[0070] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0071] Figure 1 A flowchart of a terahertz near-field radar image enhancement method provided by the present invention;

[0072] Figure 2 The images provided in this embodiment of the invention are terahertz near-field radar images to be enhanced; wherein, (a) is the first terahertz near-field radar image to be enhanced, and (b) is the second terahertz near-field radar image to be enhanced.

[0073] Figure 3 The images provided in this embodiment of the invention are enhanced terahertz near-field radar images using the method of the present invention; wherein, (a) is the first enhanced terahertz near-field radar image, and (b) is the second enhanced terahertz near-field radar image. Detailed Implementation

[0074] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0075] like Figure 1 As shown in the figure, an embodiment of the present invention discloses a method for enhancing terahertz near-field radar images, comprising the following steps:

[0076] S1: Perform data conversion and normalization on the enhanced terahertz near-field radar image to obtain a normalized terahertz near-field radar image;

[0077] In one embodiment, the data conversion in S1 converts the pixel values ​​of all pixels in the terahertz near-field radar image to be enhanced from 8-bit unsigned integers to 32-bit floating-point numbers.

[0078] The normalization in S1 linearly maps the pixel values ​​of all pixels in the enhanced terahertz near-field radar image from the interval [0, 255] to the interval [0, 1].

[0079] In one embodiment, the terahertz near-field radar image to be enhanced is acquired by an active area array imaging terahertz near-field radar imaging system. The imaging frequency range of the terahertz near-field radar imaging system is 5-10Thz, and the pixel size of the terahertz near-field radar image to be enhanced is 640*480.

[0080] It is understandable that converting the 8-bit unsigned integer pixel value to a 32-bit floating-point number can avoid the precision loss caused by subsequent division.

[0081] In one embodiment, the normalization is performed using the following formula: ; This represents the terahertz near-field radar image to be enhanced after data conversion, with the interval [0, 255]. This represents a normalized terahertz near-field radar image, with the interval [0,1].

[0082] S2: Perform stationary wavelet transform decomposition on the normalized terahertz near-field radar image to obtain several low-frequency sub-bands and several high-frequency sub-bands.

[0083] In one embodiment, S2 specifically includes the following steps:

[0084] S21: The normalized terahertz near-field radar image is decomposed into a first-level decomposition using the low-pass and high-pass filters of the Haar filter bank to obtain the low-frequency subband LL1, the horizontal high-frequency subband LH1, the vertical high-frequency subband HL1, and the diagonal high-frequency subband HH1.

[0085] S22: The low-frequency subband LL1 is decomposed in the second stage using the low-pass and high-pass filters of the Haar filter bank to obtain the low-frequency subband LL2, the horizontal high-frequency subband LH2, the vertical high-frequency subband HL2, and the diagonal high-frequency subband HH2.

[0086] In one embodiment, the calculation formulas for the low-frequency sub-band LL1, the horizontal high-frequency sub-band LH1, the vertical high-frequency sub-band HL1, the diagonal high-frequency sub-band HH1, the low-frequency sub-band LL2, the horizontal high-frequency sub-band LH2, the vertical high-frequency sub-band HL2, and the diagonal high-frequency sub-band HH2 are as follows:

[0087] ;

[0088] ;

[0089] ;

[0090] ;

[0091] ;

[0092] ;

[0093] ;

[0094] ;

[0095] in, This represents the normalized terahertz near-field radar image; and The numbers represent the low-pass and high-pass filters of the Haar filter bank, respectively; * represents convolution operation; and T represents transpose.

[0096] It is understandable that low-frequency components are used to preserve brightness information, while high-frequency components are used to capture details in different directions.

[0097] In one embodiment: ; .

[0098] S3: High-frequency subbands are filtered in multiple directions using filters of different directions to obtain several enhanced high-frequency subbands;

[0099] In one embodiment, high-frequency subbands are subjected to multi-directional high-frequency filtering using filters in different directions to obtain several enhanced high-frequency subbands, based on the following formula:

[0100] ;

[0101] ;

[0102] ;

[0103] ;

[0104] ;

[0105] ;

[0106] Where * represents the convolution operation; K r Let K1 represent the r-th directional filter, where r = 1, 2, ..., 8. K1 represents the horizontal directional filter, K2 represents the vertical directional filter, K3 represents the 45° diagonal directional filter, K4 represents the 135° diagonal directional filter, K5 represents the horizontal edge enhancement filter, K6 represents the vertical edge enhancement filter, K7 represents the Laplace filter, and K8 represents the sharpening filter. This indicates that the enhanced horizontal high-frequency subband LH1 is obtained by performing high-frequency multi-directional filtering on the horizontal high-frequency subband LH1. This indicates the enhanced vertical high-frequency subband obtained by performing high-frequency multi-directional filtering on the vertical high-frequency subband HL1. This indicates that the enhanced diagonal high-frequency subband HH1 is obtained by performing high-frequency multi-directional filtering on the diagonal high-frequency subband HH1. This indicates that the enhanced horizontal high-frequency subband LH2 is obtained by performing high-frequency multi-directional filtering on the horizontal high-frequency subband LH2. This indicates the enhanced vertical high-frequency subband obtained by performing high-frequency multi-directional filtering on the vertical high-frequency subband HL2. This indicates that the enhanced diagonal high-frequency subband HH2 is obtained by performing high-frequency multi-directional filtering.

[0107] In one embodiment: ; ; ; ; ; ; ; .

[0108] It should be noted that:

[0109] The horizontal filter K1 is used to calculate the gradient in the vertical direction of the image and enhance the brightness changes in the horizontal direction.

[0110] The vertical direction filter K2 is used to calculate the gradient in the horizontal direction of the image and enhance the brightness changes in the vertical direction;

[0111] The 45° diagonal filter K3 is used to detect edges in the 135° direction, i.e., brightness changes from the upper left to the lower right.

[0112] The 135° diagonal filter K4 is used to detect edges in the 45° direction, i.e., brightness changes from the upper right to the lower left.

[0113] The horizontal edge enhancement filter K5 enhances horizontal edges by using high weights in the center row;

[0114] The vertical edge enhancement filter K6 enhances vertical edges by using high weights in the center column;

[0115] The Laplace filter K7 enhances edges in all directions;

[0116] The sharpening filter K8 is used for image sharpening.

[0117] The mean and standard deviation of the low-frequency subband are used to perform low-frequency adaptive linear enhancement of the low-frequency subband, resulting in several enhanced low-frequency subbands.

[0118] In one embodiment, the low-frequency sub-band is subjected to low-frequency adaptive linear enhancement using the mean and standard deviation of the low-frequency sub-band to obtain several enhanced low-frequency sub-bands, based on the following formula:

[0119] ;

[0120] ;

[0121] ;

[0122] ;

[0123] ;

[0124] ;

[0125] ;

[0126] ;

[0127] ;

[0128] ;

[0129] ;

[0130] ;

[0131] ;

[0132] ;

[0133] ;

[0134] ;

[0135] Where M represents the number of rows of the low-frequency subband LL1 and the low-frequency subband LL2; N represents the number of columns of the low-frequency subband LL1 and the low-frequency subband LL2; This represents the pixel value of the pixel in the i-th row and j-th column of the low-frequency subband LL1; and These represent the mean and standard deviation of the low-frequency subband LL1, respectively. , , These represent the mean adjustment coefficient, standard deviation adjustment coefficient, and global adjustment coefficient of the low-frequency subband LL1, respectively. This represents the adaptive enhancement coefficient of the low-frequency subband LL1; This represents the default enhancement factor of the low-frequency subband LL1; This represents the total enhancement factor of the low-frequency subband LL1; This indicates that the enhanced low-frequency subband LL1 is obtained by performing low-frequency adaptive linear enhancement. This represents the pixel value of the pixel in the i-th row and j-th column of the low-frequency subband LL2; and These represent the mean and standard deviation of the low-frequency subband LL2, respectively. , , These represent the mean adjustment coefficient, standard deviation adjustment coefficient, and global adjustment coefficient of the low-frequency subband LL2, respectively. This represents the adaptive enhancement coefficient of the low-frequency subband LL2; This represents the default enhancement factor of the low-frequency subband LL2; This represents the total enhancement factor of the low-frequency subband LL2; This indicates that the low-frequency subband LL2 is subjected to low-frequency adaptive linear enhancement, resulting in the enhanced low-frequency subband.

[0136] It is understandable that: and Reflects the overall brightness of the image; and Measure low-frequency contrast;

[0137] When the image is too dark or too bright, this invention sets up a threshold-based adjustment mechanism. ( When the mean is less than a preset threshold of 0.5, positive intervention is taken to improve the mean; when ( When the mean exceeds a preset threshold of 0.5, negative intervention is applied to reduce the mean; similarly, the standard deviation is adjusted based on a set threshold of 0.2. ( Adjustments were made: standard deviation was reused. ( Adjusting the overall image contrast, setting a threshold of 0.15 to detect extremely low threshold scenes, and a coefficient of 1.2 to provide gentle adjustments:

[0138] This invention sets ( The coefficient is 0.7, meaning that adaptive control accounts for 70%, prioritizing data-driven dynamic adjustments; [This is followed by a seemingly unrelated sentence: "Set..."] ( The coefficient is 0.3, meaning that human intervention accounts for 30%, to prevent overcorrection in extreme cases and improve robustness.

[0139] S4: Perform inverse stationary wavelet transform on the several enhanced high-frequency subbands and the several enhanced low-frequency subbands to obtain the reconstructed terahertz near-field radar image;

[0140] In one embodiment, S4 is implemented based on the following formula:

[0141] ;

[0142] in, The image represents the reconstructed terahertz near-field radar image; * indicates a convolution operation.

[0143] S5: Perform a linear transformation on the reconstructed terahertz near-field radar image to obtain an enhanced terahertz near-field radar image.

[0144] In one embodiment, S5 is implemented based on the following formula:

[0145] ;

[0146] in, This refers to the enhanced terahertz near-field radar image.

[0147] In one embodiment, the above-described method for enhancing terahertz near-field radar images further includes the following steps:

[0148] S6: The enhanced terahertz near-field radar image is locally contrast optimized (enhancing local contrast and improving details in dark / bright areas) using a limited contrast histogram equalization algorithm to obtain the final enhanced terahertz near-field radar image.

[0149] In one embodiment, S6 specifically includes the following steps;

[0150] S61: Divide the enhanced terahertz near-field radar image into m×n local regions; wherein, the size of each local region is... m and n are positive integers;

[0151] S62: Calculate the gray-level histogram of the s-th local region. Where k represents the gray level; ; ;

[0152] S63: Gray-level histogram Threshold cropping is performed to obtain the cropped grayscale histogram. ;in, Indicates taking The smaller value between and 2;

[0153] S64: Calculate the cumulative distribution function of the s-th local region. ;

[0154] S65: Set pixel value Substitute In k, obtain enhanced pixels ;in, express The pixel value of the e-th pixel; The e-th pixel is located within the s-th local region; Enhanced pixels This represents the pixel value of the e-th pixel in the s-th local region of the final enhanced terahertz near-field radar image.

[0155] Finally, by Figure 2 and Figure 3 It can be seen that the present invention can enhance the contrast of the terahertz near-field radar image to be enhanced, highlight the edge features of the terahertz near-field radar image to be enhanced, and enrich the detailed information of the terahertz near-field radar image to be enhanced.

[0156] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0157] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method of enhancing a terahertz near-field radar image, characterized by, The method comprises the following steps: S1: data conversion and normalization are performed on a to-be-enhanced terahertz near-field radar image to obtain a normalized terahertz near-field radar image; S2: the normalized terahertz near-field radar image is decomposed by stationary wavelet transform to obtain a plurality of low-frequency subbands and a plurality of high-frequency subbands; S3: high-frequency multi-directional filtering is performed on the high-frequency subbands by using filters of different directions to obtain a plurality of enhanced high-frequency subbands; The low-frequency adaptive linear enhancement is performed on the low-frequency subbands by using the mean value and the standard deviation of the low-frequency subbands to obtain a plurality of enhanced low-frequency subbands; S4: inverse stationary wavelet transform is performed on the plurality of enhanced high-frequency subbands and the plurality of enhanced low-frequency subbands to obtain a reconstructed terahertz near-field radar image; S5: linear transformation is performed on the reconstructed terahertz near-field radar image to obtain an enhanced terahertz near-field radar image.

2. The method according to claim 1, wherein in S1, the pixel values of all pixels of the to-be-enhanced terahertz near-field radar image are converted from 8-bit unsigned integers to 32-bit floating-point numbers; In S1, the pixel values of all pixels of the to-be-enhanced terahertz near-field radar image after data conversion are linearly mapped from the [0, 255] interval to the [0, 1] interval. S2 specifically comprises the following steps:

3. The method of claim 1 or 2, wherein S21: the normalized terahertz near-field radar image is decomposed by a low-pass filter and a high-pass filter of a Haar filter bank to obtain a low-frequency subband LL1, a horizontal high-frequency subband LH1, a vertical high-frequency subband HL1 and a diagonal high-frequency subband HH1; S22: the low-frequency subband LL1 is decomposed by a low-pass filter and a high-pass filter of a Haar filter bank to obtain a low-frequency subband LL2, a horizontal high-frequency subband LH2, a vertical high-frequency subband HL2 and a diagonal high-frequency subband HH2. The calculation formulas of the low-frequency subband LL1, the horizontal high-frequency subband LH1, the vertical high-frequency subband HL1, the diagonal high-frequency subband HH1, the low-frequency subband LL2, the horizontal high-frequency subband LH2, the vertical high-frequency subband HL2 and the diagonal high-frequency subband HH2 are as follows:

4. The method of claim 3, wherein the method further comprises: The high-frequency multi-directional filtering is performed on the high-frequency subbands by using filters of different directions to obtain a plurality of enhanced high-frequency subbands, which is realized based on the following formula: ； ； ； ； ； ； ； ； wherein denotes the normalized terahertz near-field radar image; and denotes a low-pass filter and a high-pass filter of the Haar filter bank, respectively; * denotes a convolution operation; T denotes a transpose.

5. The method of claim 4, wherein, The low-frequency adaptive linear enhancement is performed on the low-frequency subbands by using the mean value and the standard deviation of the low-frequency subbands to obtain a plurality of enhanced low-frequency subbands, which is realized based on the following formula: ； ； ； ； ； ； wherein * represents a convolution operation; K r represents the rth direction filter, r = 1, 2, …, 8, K1 represents a horizontal direction filter, K2 represents a vertical direction filter, K3 represents a 45° diagonal direction filter, K4 represents a 135° diagonal direction filter, K5 represents a horizontal edge enhancement filter, K6 represents a vertical edge enhancement filter, K7 represents a Laplacian filter, and K8 represents a sharpening filter; represents high-frequency multi-directional filtering on the horizontal high-frequency subband LH1 to obtain an enhanced horizontal high-frequency subband; represents high-frequency multi-directional filtering on the vertical high-frequency subband HL1 to obtain an enhanced vertical high-frequency subband; represents high-frequency multi-directional filtering on the diagonal high-frequency subband HH1 to obtain an enhanced diagonal high-frequency subband; represents high-frequency multi-directional filtering on the horizontal high-frequency subband LH2 to obtain an enhanced horizontal high-frequency subband; represents high-frequency multi-directional filtering on the vertical high-frequency subband HL2 to obtain an enhanced vertical high-frequency subband; represents high-frequency multi-directional filtering on the diagonal high-frequency subband HH2 to obtain an enhanced diagonal high-frequency subband.

6. The method of claim 5, wherein the method further comprises: S4 is realized based on the following formula: ； ； ； ； ； ； ； ； ； ； ； ； ； ； ； ； wherein M represents the number of rows of the low-frequency sub-band LL1 and the low-frequency sub-band LL2; N represents the number of columns of the low-frequency sub-band LL1 and the low-frequency sub-band LL2; represents the pixel value of the pixel in the i-th row and the j-th column of the low-frequency sub-band LL1; and respectively represent the mean and the standard deviation of the low-frequency sub-band LL1; , , respectively represent the mean adjustment coefficient, the standard deviation adjustment coefficient and the global adjustment coefficient of the low-frequency sub-band LL1; represents the adaptive enhancement coefficient of the low-frequency sub-band LL1; represents the default enhancement coefficient of the low-frequency sub-band LL1; represents the total enhancement coefficient of the low-frequency sub-band LL1; represents the low-frequency adaptive linear enhancement on the low-frequency sub-band LL1, and the enhanced low-frequency sub-band obtained thereby; represents the pixel value of the pixel in the i-th row and the j-th column of the low-frequency sub-band LL2; and respectively represent the mean and the standard deviation of the low-frequency sub-band LL2; , , respectively represent the mean adjustment coefficient, the standard deviation adjustment coefficient and the global adjustment coefficient of the low-frequency sub-band LL2; represents the adaptive enhancement coefficient of the low-frequency sub-band LL2; represents the default enhancement coefficient of the low-frequency sub-band LL2; represents the total enhancement coefficient of the low-frequency sub-band LL2; represents the low-frequency adaptive linear enhancement on the low-frequency sub-band LL2, and the enhanced low-frequency sub-band obtained thereby.

7. The method of claim 6, wherein the method further comprises: S5 is realized based on the following formula: ； wherein denotes the reconstructed terahertz near-field radar image; denotes a convolution operation.

8. The method of claim 7, wherein the method further comprises: The method further comprises the following steps: ； wherein represents the enhanced terahertz near-field radar image.

9. The method of claim 8, wherein, S6: a limited contrast histogram equalization algorithm is used to perform local contrast optimization on the enhanced terahertz near-field radar image to obtain a final enhanced terahertz near-field radar image. S6 specifically comprises the following steps; 10. The method of claim 9, wherein, ​ S61: dividing the enhanced terahertz near-field radar image into m x n local regions; wherein the size of each local region is ; m and n are positive integers; S62: Calculate the gray level histogram of the s-th local region ; wherein k represents the gray level; ; ; S63: thresholding the gray scale histogram to obtain a thresholded gray scale histogram ; wherein, represents the smaller value between and 2; S64: Calculate the cumulative distribution function of the s-th local region ; S65: the pixel value is brought into with k, obtaining the enhanced pixel ; wherein denotes a pixel value of the e-th pixel in the e-th pixel in ; the enhanced pixel denotes a pixel value of the e-th pixel of the s-th local region in the final enhanced terahertz near-field radar image.

Images