NSCT domain sonar image denoising method and device based on neutral set and bilateral filtering

Abstract

The application discloses a NSCT domain sonar image denoising method and device based on neutral set and bilateral filtering, and belongs to the technical field of sonar image denoising. In order to solve the problems of poor denoising effect, insufficient edge preservation ability and low operation efficiency of the existing image denoising method, the application first decomposes a noisy image based on a non-subsampled contourlet transform, then converts high-frequency subband images to a neutral set domain, uses a true subset, a false set and an uncertain set for representation, performs beta enhancement on the image of the true subset, and performs bilateral filtering processing on the enhanced true subset image; meanwhile, the low-frequency subband image is converted to the neutral set domain, the true subset image is extracted as a guide image of joint bilateral filtering, and the joint bilateral filtering is guided to perform smoothing processing on the low-frequency subband image; the NSCT inverse transform is used to reconstruct the high-frequency subband image after the bilateral filtering processing and the low-frequency subband image after the joint bilateral filtering processing, so that the finally denoised image is obtained.

Description

Technical Field

[0001] This invention belongs to the field of sonar image denoising technology, specifically relating to a sonar image denoising method and device. Background Technology

[0002] Sonar is the primary means of transmitting ocean information. However, due to the influence of sea winds, ocean currents, water temperature, impurities, and imaging equipment, images obtained using sonar detection technology are subject to far more severe noise interference than optical images. Sonar images typically contain various types of noise, primarily manifesting as granular speckle noise. This speckle noise is multiplicative, thus severely degrading image quality, causing blurring, loss of edge details, and negatively impacting subsequent image segmentation and recognition.

[0003] In recent years, commonly used algorithms for sonar image denoising include wavelet transform and contourlet transform. However, wavelet denoising cannot effectively capture edge and contour features, ridge transform can only represent linear singular features and cannot effectively describe the curve singular features in the signal; curvelet transform has a complex structure and high computational cost. These methods result in the inaccurate capture of image details and low denoising efficiency. Currently, there are various optical image denoising algorithms, such as partial differential equation denoising, total variation denoising, and mathematical morphology denoising, which ignore the inherent characteristics of the pixels themselves, leading to image blurring or loss of detail.

[0004] In summary, existing sonar image denoising methods suffer from problems such as insignificant denoising effect, poor edge preservation ability, and low computational efficiency. Therefore, how to effectively remove noise from sonar images under severe noise interference, improve edge preservation ability, and enhance computational efficiency is an urgent problem to be solved. Summary of the Invention

[0005] In order to solve the problems of poor denoising effect, insufficient edge preservation ability and low computational efficiency of existing image denoising methods, this invention proposes a sonar image denoising method based on neutral set and bilateral filtering in the NSCT domain.

[0006] A method for denoising NSCT domain sonar images based on neutral set and bilateral filtering includes the following steps:

[0007] Step 1: For underwater sonar images, use non-subsampled contour wave transform (NSCT) to decompose the noisy images and obtain multiple high-frequency subband images and one low-frequency subband image.

[0008] Step 2: Using a clustering method based on neutral sets, each high-frequency subband image is clustered, that is, the high-frequency subband image is transformed into the neutral set domain and represented by proper subsets, false sets, and uncertain sets;

[0009] Beta enhancement is performed on the image T(x,y) of the proper subset to obtain the enhanced image T of the proper subset. β ';

[0010] Step 3: For the enhanced images of the true subset, false set, and uncertain set obtained in Step 2, perform bilateral filtering on the enhanced true subset image; perform zero-value processing on the false set and uncertain set images, so they will not participate in the inverse NSCT transform.

[0011] Step 4: Convert the low-frequency subband image to the neutral set domain as well, and extract the true subset image as the guide map for joint bilateral filtering to guide the joint bilateral filtering to smooth the low-frequency subband image;

[0012] Step 5: Reconstruct the high-frequency subband image after bilateral filtering in Step 3 and the low-frequency subband image after joint bilateral filtering in Step 4 using the inverse NSCT transform to obtain the final denoised image.

[0013] Furthermore, the high-frequency subband image is transformed to the neutral set domain, expressed using the tables of proper subsets, false sets, and uncertain sets as follows:

[0014]

[0015] F(x,y)=1-T(x,y)

[0016]

[0017]

[0018] In the formula, G(x,y) is the pixel value at the position (x,y) of the input image pixel. The image is after median filtering, min and max represent the minimum and maximum values, respectively; T(x,y) represents the image of the proper subset, F(x,y) represents the image of the false set, and I(x,y) represents the image of the uncertain set; δ(x,y) is the intermediate variable, and abs(·) represents the function for calculating the absolute value.

[0019] Furthermore, the image of the enhanced proper subset

[0020] Furthermore, step three involves performing bilateral filtering on the enhanced proper subset image as follows:

[0021]

[0022]

[0023]

[0024] In the formula, ws (a,b), w g (a, b) represent the spatial weight and gray-level similarity weight, respectively, σ s σ g Let Z be the spatial standard deviation and the grayscale standard deviation, respectively; Z be the filter size; and G′(x,y) be the input image T. β G′(a,b) is the pixel value at pixel position (x,y), and G′(a,b) is the pixel value at the center pixel position (a,b) of the input image. This is the result of bilateral filtering of G′(x,y).

[0025] Furthermore, the specific process of step one includes the following steps:

[0026] Step 1: Perform an upsampling operation on the Laplace pyramid filter to construct a dual-channel non-downsampling pyramid filter, i.e., NSPFB;

[0027] Steps 1 and 2: Use a non-subsampled pyramid filter to divide the sonar image into a high-frequency subband and a low-frequency subband;

[0028] Step 13: Using a non-subsampled directional filter (NSDFB), the high-frequency subband is further decomposed to obtain multiple directional bands of the high-frequency subband; the l-th layer can obtain 2 l A total of [number] high-frequency subband images can be obtained. A high-frequency subband image.

[0029] Furthermore, the sampling matrix of the dual-channel non-downsampled pyramid in step one is: and

[0030] Furthermore, in steps one and two, a non-downsampled pyramid filter is used to decompose the noisy image into two layers to obtain the low-frequency and high-frequency components of the source image, resulting in one low-frequency sub-band image and three high-frequency sub-band images.

[0031] A computer storage medium storing at least one instruction, which is loaded and executed by a processor to implement the NSCT domain sonar image denoising method based on neutral set and bilateral filtering.

[0032] A sonar image denoising device based on neutral set and bilateral filtering in the NSCT domain, the device comprising a processor and a memory, wherein the memory stores at least one instruction, the at least one instruction being loaded and executed by the processor to implement the sonar image denoising method based on neutral set and bilateral filtering in the NSCT domain.

[0033] The beneficial effects of this invention are:

[0034] This invention combines neutral set theory, bilateral filtering, and NSCT to maximize the denoising effect and edge preservation of sonar images. It also addresses the problems of low contrast and severe noise interference in sonar images. The invention decomposes the noisy image using NSCT, then applies neutral set theory to remove noise signals from high-frequency coefficients while preserving detail signals. Low-frequency coefficients are then subjected to joint bilateral filtering to smooth the image and improve contrast. Finally, an inverse NSCT transform is performed to obtain the denoised image. Furthermore, this invention effectively solves the problem of low computational efficiency. Attached Figure Description

[0035] Figure 1 This is a flowchart of NSCT domain sonar image denoising based on neutral set and bilateral filtering.

[0036] Figure 2 This is a schematic diagram of NSCT decomposition.

[0037] Figure 3 Image of the shipwreck wreckage before noise reduction.

[0038] Figure 4 This is an image of the shipwreck after noise reduction. Detailed Implementation

[0039] Specific implementation method one: Combining Figure 1 This implementation method is described below.

[0040] The NSCT domain sonar image denoising method based on neutral set and bilateral filtering described in this embodiment specifically includes the following steps:

[0041] Step 1: For underwater sonar images, decompose the noisy images using non-subsampled contourlet transform (NSCT) to obtain multiple high-frequency subband images and one low-frequency subband image.

[0042] In contour wave transform (CT), NSCT eliminates the direct sampling operation of the image and allows the corresponding filters to perform sampling operations. During decomposition, it uses the non-subsampled pyramid filter (NSPFB) for multi-scale decomposition and uses the non-subsampled directional filter (NSDFB) for the high-frequency subbands of multi-scale decomposition. NSCT eliminates the downsampling operation in the Laplacian pyramid filter and instead uses upsampling to complete the dual-channel non-subsampled pyramid.

[0043] The specific process of step one includes the following steps:

[0044] Step 11: Construct a dual-channel non-downsampled pyramid filter (NSPFB) by upsampling the Laplace pyramid filter; the sampling matrix of the dual-channel NSPFB is as follows. and

[0045] Steps 1 and 2: Use a non-subsampled pyramid filter to divide the sonar image into a high-frequency subband and a low-frequency subband;

[0046] In this embodiment, a non-subsampled pyramid filter is used to decompose the noisy image into two layers to obtain the low-frequency and high-frequency components of the source image, resulting in one low-frequency sub-band image and three high-frequency sub-band images.

[0047] Step 13: Further decompose the high-frequency subband using a non-subsampled directional filter (NSDFB) to obtain multiple directional bands for the high-frequency subband; the l-th layer can obtain 2 l A total of [number] high-frequency subband images can be obtained. A high-frequency sub-band image;

[0048] Step 2: Using a clustering method based on neutral sets, each high-frequency sub-band image is clustered to divide the signal into detail signals and noise signals; the specific process includes the following steps:

[0049] The high-frequency subband image is transformed to the neutral set domain, represented by proper subsets, false sets, and uncertain sets, and its expression is:

[0050]

[0051] F(x,y)=1-T(x,y)

[0052]

[0053]

[0054] In the formula, G(x,y) is the pixel value at the position (x,y) of the input image pixel. The image is after median filtering, min and max represent the minimum and maximum values, respectively; T(x,y) represents the image of the proper subset, F(x,y) represents the image of the false set, and I(x,y) represents the image of the uncertain set; δ(x,y) is the intermediate variable, and abs(·) represents the function for calculating the absolute value.

[0055] The image of the proper subset is beta-enhanced to obtain the enhanced image of the proper subset. The beta enhancement is as follows:

[0056]

[0057] Beta enhancement can reduce uncertainty in the computation process, making the deterministic set clearer and with higher contrast;

[0058] The true subset obtained by the above process contains the main information of the image, the false set is noise, and the uncertain set is noise and background interference. The main information in the high-frequency subband image is the image edge, thus achieving the purpose of separating the effective signal from the noise signal.

[0059] Step 3: For the enhanced images of the true subset, false set, and uncertain set obtained in Step 2, perform bilateral filtering on the enhanced true subset image; perform zero-value processing on the false set and uncertain set images, so they will not participate in the inverse NSCT transform.

[0060] Bilateral filtering allows for setting the spatial and grayscale standard deviations as needed, achieving noise reduction while preserving edges; the expression is as follows:

[0061]

[0062]

[0063]

[0064] In the formula, w s (a,b), w g (a, b) represent the spatial weight and gray-level similarity weight, respectively, σ s σ g Let S be the spatial standard deviation and the grayscale standard deviation, respectively; Z be the filter size; and G′(x,y) be the input image T′. β The pixel value at pixel position (x, y), and G′(a, b) is the pixel value at the center pixel position (a, b) of the input image. This is the result of bilateral filtering of G′(x,y).

[0065] Step 4: Convert the low-frequency subband image to the neutral set domain as well, and extract the true subset image as the guide map for joint bilateral filtering to guide the joint bilateral filtering to smooth the low-frequency subband image;

[0066] The calculation method of joint bilateral filtering is the same as that of bilateral filtering. The joint bilateral filter introduces a guiding image on the basis of bilateral filtering, which makes the weights more stable.

[0067] Step 5: Reconstruct the high-frequency subband image after bilateral filtering in Step 3 and the low-frequency subband image after joint bilateral filtering in Step 4 using the inverse NSCT transform to obtain the final denoised image.

[0068] Example

[0069] This embodiment of an underwater sonar image denoising method based on image processing includes: firstly, decomposing the acquired sonar image using NSCT, processing the high-frequency coefficients using neutral set to remove noise signals and retain detail signals; secondly, performing bilateral filtering on the low-frequency coefficients to smooth the image and improve the visual effect; and finally reconstructing the image using inverse NSCT to obtain the denoised sonar image.

[0070] A denoising experiment was conducted using images of shipwreck wreckage as an example. The specific process is as follows:

[0071] Step 1: Use NSCT to decompose the noisy na image to obtain high-frequency subband images and low-frequency subband images.

[0072] Step 1: Perform an upsampling operation on the Laplace pyramid filter to construct a two-channel non-downsampling pyramid filter;

[0073] Step 2: Use a non-subsampled pyramid filter to divide the sonar image into a high-frequency subband and a low-frequency subband;

[0074] Step 3: Use the non-subsampled directional filter (NSDFB) to further decompose the high-frequency subband to obtain multiple directional bands of the high-frequency subband.

[0075] Step 2: Convert each high-frequency subband image to the neutral set domain.

[0076] Step 3: Based on the detail signal and noise signal obtained in Step 2, perform bilateral filtering and zero-value processing respectively.

[0077] Step 4: Obtain the guiding graph of the joint bilateral filter to guide the smoothing and noise reduction of low-frequency coefficients; the specific process is as follows:

[0078] Step 1: Transform the low-frequency coefficients to the neutral set domain;

[0079] Step 2: Extract the proper subset from Step 1 as the guide graph;

[0080] Step 3: Use the guide graph from Step 2 to guide the joint bilateral filtering to smooth and denoise the low-frequency coefficients, reducing the parameter settings for spatial standard deviation and grayscale standard deviation.

[0081] Step 5: Reconstruct the high-frequency subband and low-frequency subband images processed in steps 2 to 4 using the inverse NSCT transform.

[0082] Noisy images of shipwreck debris, such as Figure 3 As shown, the image suffers from severe noise interference and low contrast, affecting its overall quality and visual appeal. After processing by this invention, the image is as follows: Figure 4The image shown is the denoised image. It is evident that this invention improves the overall image quality and visual appeal of underwater shipwreck images by denoising them, enhancing edge preservation capabilities and reducing computational load. Specific Implementation Method Two:

[0084] This embodiment is a computer storage medium that stores at least one instruction. The at least one instruction is loaded and executed by a processor to implement the NSCT domain sonar image denoising method based on neutral set and bilateral filtering.

[0085] It should be understood that the instructions include computer program products, software, or computerized methods corresponding to any method described in this invention; the instructions can be used to program computer systems or other electronic devices. Computer storage media may include readable media on which instructions are stored, and may include, but are not limited to, magnetic storage media, optical storage media; magneto-optical storage media include read-only memory (ROM), random access memory (RAM), erasable programmable memory (e.g., EPROM and EEPROM), and flash memory layers, or other types of media suitable for storing electronic instructions. Specific implementation method three:

[0087] This embodiment is an NSCT domain sonar image denoising device based on neutral set and bilateral filtering. The device includes a processor and a memory. It should be understood that this includes any device including a processor and a memory described in this invention. The device may also include other units and modules that perform display, interaction, processing, control and other functions through signals or instructions.

[0088] The memory stores at least one instruction, which is loaded and executed by the processor to implement the NSCT domain sonar image denoising method based on neutral set and bilateral filtering.

[0089] The above examples of the present invention are merely illustrative of the computational model and process of the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is impossible to exhaustively list all possible implementations here. Any obvious variations or modifications derived from the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for denoising NSCT domain sonar images based on neutral set and bilateral filtering, characterized in that, Includes the following steps: Step 1: For underwater sonar images, decompose the noisy image using non-subsampled contourlet transform (NSCT) to obtain multiple high-frequency sub-band images and one low-frequency sub-band image; including the following steps: Step 11: Perform an upsampling operation on the Laplace pyramid filter to construct a dual-channel non-downsampling pyramid filter, i.e., NSPFB; the sampling matrix of the dual-channel non-downsampling pyramid is as follows: and ; Step 1 and 2: Use a non-subsampled pyramid filter to divide the sonar image into a high-frequency sub-band and a low-frequency sub-band; use a non-subsampled pyramid filter to decompose the noisy image into two layers to obtain the low-frequency and high-frequency components of the source image, resulting in one low-frequency sub-band image and three high-frequency sub-band images. Step 13: Using a non-subsampled directional filter (NSDFB), further decompose the high-frequency subband to obtain multiple directional bands of the high-frequency subband; Layers can be obtained A total of [number] high-frequency subband images can be obtained. A high-frequency sub-band image; Step 2: Using a clustering method based on neutral sets, each high-frequency subband image is clustered, that is, the high-frequency subband image is transformed into the neutral set domain and represented by proper subsets, false sets, and uncertain sets; Images of proper subsets conduct The image of the enhanced proper subset is obtained. The image of the enhanced proper subset ; Step 3: For the enhanced images of the true subset, false set, and uncertain set obtained in Step 2, perform bilateral filtering on the enhanced true subset image; perform zero-value processing on the false set and uncertain set images, so they will not participate in the inverse NSCT transform. Step 4: Convert the low-frequency subband image to the neutral set domain as well, and extract the true subset image as the guide map for joint bilateral filtering to guide the joint bilateral filtering to smooth the low-frequency subband image; Step 5: Reconstruct the high-frequency subband image after bilateral filtering in Step 3 and the low-frequency subband image after joint bilateral filtering in Step 4 using the inverse NSCT transform to obtain the final denoised image.

2. The NSCT domain sonar image denoising method based on neutral set and bilateral filtering according to claim 1, characterized in that, The high-frequency subband image is transformed to the neutral set domain, represented by proper subsets, false sets, and uncertain sets, as shown in the following expression: In the formula, Given the pixel value at the (x, y) position of the input image pixel. The image is after median filtering, where min and max represent the minimum and maximum values, respectively. Images representing proper subsets, The image representing the pseudo set, An image representing an uncertain set; Let abs be an intermediate variable, and abs(·) represents a function to calculate the absolute value.

3. The NSCT domain sonar image denoising method based on neutral set and bilateral filtering according to claim 2, characterized in that, Step three involves performing bilateral filtering on the enhanced proper subset image as follows: In the formula, , These are spatial weights and gray-level similarity weights, respectively. , These are the spatial standard deviation and the grayscale standard deviation, respectively. This is the filter size; For the input image The pixel value at pixel position (x, y). Given the pixel value at the center pixel position (a, b) of the input image, for The bilateral filtering results.

4. A computer storage medium, characterized in that, The storage medium stores at least one instruction, which is loaded and executed by a processor to implement the NSCT domain sonar image denoising method based on neutral set and bilateral filtering as described in any one of claims 1 to 3.

5. A sonar image denoising device based on neutral set and bilateral filtering in the NSCT domain, characterized in that, The device includes a processor and a memory, the memory storing at least one instruction, which is loaded and executed by the processor to implement the NSCT domain sonar image denoising method based on neutral set and bilateral filtering as described in any one of claims 1 to 3.

Images