A hierarchical image enhancement method based on four-orientation neighborhood edge-preserving denoising operator

Abstract

The application provides a layered image enhancement method based on a four-orientation neighborhood edge-preserving denoising operator, and belongs to the technical field of image processing. The method first realizes image adaptive layering by adaptive regularization parameter calculation and guided filter improvement based on a Sobel operator to obtain a base layer and a detail layer; secondly, a four-orientation neighborhood edge-preserving denoising operator is designed, a plurality of direction-sensitive 3*3 operators are fused into a single 5*5 composite convolution operator, and the detail layer is subjected to edge-preserving denoising treatment; finally, the base layer is subjected to contrast enhancement and fused with the processed detail layer to output an enhanced image. The application can effectively suppress noise while maintaining clear edges, and improve the overall quality and visual effect of an image.

Description

Technical Field

[0001] This invention relates to the field of image enhancement and signal processing technology, and in particular to a layered image enhancement method based on a four-way neighborhood edge-preserving denoising operator. Background Technology

[0002] Image enhancement methods based on hierarchical strategies typically decompose an image into a basal layer and a detail layer, performing contrast enhancement and detail optimization separately, and finally fusing them to improve visual effects. Guided filtering is one of the commonly used hierarchical methods, with good edge-preserving properties, but its parameter settings depend on experience, its adaptive ability is limited, and it is prone to noise residue or detail loss in different scenarios.

[0003] Detail layer processing is crucial to the enhancement effect. Traditional methods often employ linear stretching or ordinary filtering, but these often amplify noise while enhancing details or blur edges during denoising. Some existing methods remove detail layer noise through masking, such as "Adaptive Infrared Image Enhancement Algorithm Based on Guided Filter" (Tianjin Institute of Aeronautical Technology and Physics. Adaptive Infrared Image Enhancement Algorithm Based on Guided Filter: 202211148584.8[P]. 2022-12-30.), or improve performance through multi-operator collaborative processing, such as "Infrared Image Enhancement Based on Adaptive Guided Filter" (Zhang, Hui, et al. "Infrared image enhancement basedon adaptive guided filter and global–local mapping." Photonics. Vol. 11. No.8. MDPI, 2024.). However, these methods require separate denoising and edge preservation calculations, resulting in a large number of operators and frequent convolutions, leading to high computational complexity and poor real-time performance.

[0004] Therefore, how to effectively preserve edge information while ensuring denoising performance and reducing computational complexity has become a technical problem that needs to be solved in this field. Summary of the Invention

[0005] This invention proposes a hierarchical image enhancement method based on a four-way neighborhood edge-preserving denoising operator to solve the technical problems of insufficient adaptive ability, difficulty in balancing detail layer denoising and edge preservation, and high computational complexity in existing hierarchical enhancement methods.

[0006] This invention is achieved through the following technical solution: This invention proposes a layered image enhancement method based on a four-way neighborhood edge-preserving denoising operator, the method comprising: Step 1: Input the grayscale image to be processed; Step 2: Calculate the adaptive regularization parameters for the input image data. For each local window r in the image, calculate the median pixel value, and then calculate the variance of each pixel relative to that median. The variance matrix is ​​obtained; after sorting the variance matrix, the extreme values ​​at both ends are truncated, and the mean of the remaining part is taken. As the global comprehensive robust variance, the final regularization parameter is obtained through exponential mapping. ; Step 3: Use a guided filtering method based on the Sobel operator to perform image layering to obtain the base layer image; Step 4: Subtract the original image from the base layer image to obtain the detail layer image; Step 5: Construct a four-way neighborhood edge-preserving denoising operator and process the detail layer; Step 6: Perform contrast-limited adaptive histogram equalization on the base layer image to improve its contrast. Step 7: The processed base layer and detail layer are fused using linear weighting to obtain the final enhanced image.

[0007] Furthermore, the grayscale image contains low-frequency background, high-frequency details, and noise information.

[0008] Further, in step 3, the edge gradient magnitude of the image is calculated. G According to the preset gradient threshold Distinguish between edge and non-edge regions; decrease the regularization parameter in edge regions to enhance edge preservation, and increase the regularization parameter in non-edge regions to enhance noise suppression; perform guided filtering to obtain the base layer image.

[0009] Furthermore, in step 5, neighborhood edge-preserving denoising operators corresponding to different directions are constructed and fused into a composite convolution operator; this operator is then used to perform convolution calculations on the detail layers. R An adaptive threshold is set based on the mean and variance of the grayscale of the detail layer. The pixel is determined to be an edge or noise based on the comparison result between the convolution value and the threshold. The noise pixels are then set to zero to obtain the denoised detail layer.

[0010] Furthermore, in step 2, the truncation ratio of the extreme values ​​at both ends of the variance matrix array is determined according to requirements to eliminate the influence of outliers.

[0011] Furthermore, the gradient threshold mentioned in step 3 The threshold is adjusted based on the actual image content; the smaller the threshold, the more areas are identified as edges.

[0012] Furthermore, in step 5, nine 3×3 neighborhood edge-preserving and denoising operators corresponding to different directions are constructed and fused into a 5×5 composite convolution operator.

[0013] Furthermore, the nine 3×3 operators mentioned in step 5 are merged into a 5×5 convolution operator by superimposing the weights at corresponding positions.

[0014] The present invention also proposes an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the layered image enhancement method based on a four-way neighborhood edge-preserving denoising operator.

[0015] The present invention also proposes a computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the steps of the layered image enhancement method based on a four-way neighborhood edge-preserving denoising operator.

[0016] The beneficial effects of this invention are: 1. By using adaptive regularization parameters and Sobel operator-guided filtering layering, the adaptive processing capability for images with different content is improved, maintaining clear edges while smoothing noise; 2. The designed multi-directional edge-preserving and denoising fusion operator can achieve both denoising and edge preservation simultaneously with only one convolution, significantly reducing computational complexity and improving the real-time performance of the algorithm; 3. The detail layer employs a dynamic threshold decision mechanism, which enhances the ability to distinguish between noise and edges and avoids loss of detail; 4. The overall approach effectively suppresses noise while improving image contrast and detail, resulting in improved visual quality and objective evaluation metrics. Attached Figure Description

[0017] 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.

[0018] Figure 1 This is an overall flowchart of the layered image enhancement method based on the four-way neighborhood edge-preserving denoising operator described in this invention; Figure 2 A schematic diagram of the four-way neighborhood edge-preserving and noise-reducing operator structure; Figure 3 A comparison chart showing the effects of the improved guided filtering method versus the original method; Figure 4A comparison chart showing the effects of edge-preserving and noise reduction methods for detail layers compared to traditional methods; Figure 5 A comparison chart of evaluation metrics for guided filtering methods in different scenarios; Figure 6 A comparison chart of evaluation metrics for detail layer processing methods in different scenarios. Detailed Implementation

[0019] 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.

[0020] This invention proposes a layered image enhancement method based on a four-directional neighborhood edge-preserving denoising operator, belonging to the field of image processing technology. The method first achieves adaptive image layering through adaptive regularization parameter calculation and guided filtering based on the Sobel operator, obtaining a base layer and a detail layer. Second, a four-directional neighborhood edge-preserving denoising operator is designed, fusing multiple directionally sensitive 3×3 operators into a single 5×5 composite convolution operator to perform edge-preserving denoising on the detail layer. Finally, the base layer undergoes contrast enhancement and is fused with the processed detail layer to output an enhanced image. This invention can effectively suppress noise while maintaining clear edges, improving the overall image quality and visual effect.

[0021] Specifically, in combination Figures 1-6 This invention proposes a hierarchical image enhancement method based on a four-way neighborhood edge-preserving denoising operator, the method comprising: Step 1: Input the grayscale image to be processed; Step 2: Calculate the adaptive regularization parameters for the input image data. For each local window r in the image, calculate the median pixel value, and then calculate the variance of each pixel relative to that median. The variance matrix is ​​obtained; after sorting the variance matrix, the extreme values ​​at both ends are truncated, and the mean of the remaining part is taken. As the global comprehensive robust variance, the final regularization parameter is obtained through exponential mapping. ; Step 3: Use a guided filtering method based on the Sobel operator to perform image layering to obtain the base layer image; Step 4: Subtract the original image from the base layer image to obtain the detail layer image; Step 5: Construct a four-way neighborhood edge-preserving denoising operator and process the detail layer; Step 6: Perform contrast-limited adaptive histogram equalization on the base layer image to improve its contrast. Step 7: The processed base layer and detail layer are fused using linear weighting to obtain the final enhanced image.

[0022] Furthermore, the grayscale image contains low-frequency background, high-frequency details, and noise information.

[0023] Further, in step 3, the edge gradient magnitude of the image is calculated. G According to the preset gradient threshold Distinguish between edge and non-edge regions; decrease the regularization parameter in edge regions to enhance edge preservation, and increase the regularization parameter in non-edge regions to enhance noise suppression; perform guided filtering to obtain the base layer image.

[0024] Furthermore, in step 5, neighborhood edge-preserving denoising operators corresponding to different directions are constructed and fused into a composite convolution operator; this operator is then used to perform convolution calculations on the detail layers. R An adaptive threshold is set based on the mean and variance of the grayscale of the detail layer. The pixel is determined to be an edge or noise based on the comparison result between the convolution value and the threshold. The noise pixels are then set to zero to obtain the denoised detail layer.

[0025] Furthermore, in step 2, the truncation ratio of the extreme values ​​at both ends of the variance matrix array is determined according to requirements to eliminate the influence of outliers.

[0026] Furthermore, the gradient threshold mentioned in step 3 The threshold is adjusted based on the actual image content; the smaller the threshold, the more areas are identified as edges.

[0027] Furthermore, in step 5, nine 3×3 neighborhood edge-preserving and denoising operators corresponding to different directions are constructed and fused into a 5×5 composite convolution operator.

[0028] Furthermore, the nine 3×3 operators mentioned in step 5 are merged into a 5×5 convolution operator by superimposing the weights at corresponding positions.

[0029] Example like Figure 1 As shown, this embodiment provides a layered image enhancement method based on a four-way neighborhood edge-preserving denoising operator. Taking a 1280×1024 grayscale image as an example, the implementation steps are as follows: Step 1: Input the grayscale image to be processed.

[0030] Step 2: Calculate the adaptive regularization parameter for the input image data. Take a local window size r, calculate the median and corresponding variance of pixel values ​​in each window, sort them, remove the top and bottom 5% extreme values, calculate the mean of the remaining part as the global robust variance, and then obtain the regularization parameter ε through exponential mapping; Step 3: Calculate the image gradient magnitude using the Sobel operator and set the gradient threshold. =150 is divided into edge and non-edge regions. In the edge region, ε is decreased by 0.02, and in the non-edge region, it is increased by 0.02. Guided filtering is then performed to obtain the base layer.

[0031] Step 4: Subtract the base layer from the original image to obtain the detail layer.

[0032] Step 5: Construct four-directional neighborhood edge-preserving denoising operators. First, construct nine 3×3 neighborhood edge-preserving denoising operators in nine different directions: top left, top, top right, left, center, right, bottom left, bottom, and bottom right. Stack these nine 3×3 operators into a 5×5 composite convolution kernel and perform convolution on the detail layer. Then, based on the mean value of the detail layer... Adjusting the threshold for variance Pixels with convolution values ​​above the threshold are retained, while those below the threshold are set to zero.

[0033] Step 6: Perform adaptive histogram equalization (CLAHE) on the base layer to improve contrast.

[0034] Step 7: Linearly fuse the processed base layer and detail layer with weights α=0.7 and β=0.3 to output the final image.

[0035] Experimental results show that the method of the present invention is superior to traditional methods in terms of noise suppression, edge preservation and overall visual effect, and has high processing efficiency, making it suitable for real-time image enhancement scenarios.

[0036] The present invention also proposes an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the layered image enhancement method based on a four-way neighborhood edge-preserving denoising operator.

[0037] The present invention also proposes a computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the steps of the layered image enhancement method based on a four-way neighborhood edge-preserving denoising operator.

[0038] The memory in this application embodiment can be volatile memory or non-volatile memory, or it can include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM). It should be noted that the memory used in the methods described in this invention is intended to include, but is not limited to, these and any other suitable types of memory.

[0039] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., high-density digital video discs (DVDs)), or semiconductor media (e.g., solid-state disks (SSDs)).

[0040] In implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software. The steps of the method disclosed in the embodiments of this application can be directly implemented by a hardware processor, or by a combination of hardware and software modules in the processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, detailed descriptions are omitted here.

[0041] It should be noted that the processor in the embodiments of this application can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiments can be completed by the integrated logic circuitry in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied as execution by a hardware decoding processor, or as a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above methods.

[0042] The above provides a detailed description of the layered image enhancement method based on the four-way neighborhood edge-preserving denoising operator proposed in this invention. Specific examples have been used to illustrate the principle and implementation of this invention. The description of the above embodiments is only for the purpose of helping to understand the method and core idea of ​​this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation and application scope based on the idea of ​​this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A hierarchical image enhancement method based on a four-way neighborhood edge-preserving denoising operator, characterized in that, The method includes: Step 1: Input the grayscale image to be processed; Step 2: Calculate the adaptive regularization parameters for the input image data. For each local window r in the image, calculate the median pixel value, and then calculate the variance of each pixel relative to that median. The variance matrix is ​​obtained; after sorting the variance matrix, the extreme values ​​at both ends are truncated, and the mean of the remaining part is taken. As the global comprehensive robust variance, the final regularization parameter is obtained through exponential mapping. ; Step 3: Use a guided filtering method based on the Sobel operator to perform image layering to obtain the base layer image; Step 4: Subtract the original image from the base layer image to obtain the detail layer image; Step 5: Construct a four-way neighborhood edge-preserving denoising operator and process the detail layer; Step 6: Perform contrast-limited adaptive histogram equalization on the base layer image to improve its contrast. Step 7: The processed base layer and detail layer are fused using linear weighting to obtain the final enhanced image.

2. The method according to claim 1, characterized in that, This grayscale image contains low-frequency background, high-frequency details, and noise information.

3. The method according to claim 2, characterized in that, In step 3, the edge gradient magnitude of the image is calculated. G According to the preset gradient threshold Distinguish between edge and non-edge regions; decrease the regularization parameter in edge regions to enhance edge preservation, and increase the regularization parameter in non-edge regions to enhance noise suppression; perform guided filtering to obtain the base layer image.

4. The method according to claim 3, characterized in that, In step 5, neighborhood edge-preserving denoising operators corresponding to different directions are constructed and fused into a composite convolution operator; this operator is then used to perform convolution calculations on the detail layers. R An adaptive threshold is set based on the mean and variance of the grayscale of the detail layer. The pixel is determined to be an edge or noise based on the comparison result between the convolution value and the threshold. The noise pixels are then set to zero to obtain the denoised detail layer.

5. The method according to claim 1, characterized in that, In step 2, the truncation ratio of the extreme values ​​at both ends of the variance matrix array is determined according to the requirements to eliminate the influence of outliers.

6. The method according to claim 3, characterized in that, The gradient threshold mentioned in step 3 The threshold is adjusted based on the actual image content; the smaller the threshold, the more areas are identified as edges.

7. The method according to claim 4, characterized in that, In step 5, nine 3×3 neighborhood edge-preserving and denoising operators corresponding to different directions are constructed and merged into a 5×5 composite convolution operator.

8. The method according to claim 7, characterized in that, In step 5, the nine 3×3 operators are merged into a single 5×5 convolution operator by superimposing the weights at corresponding positions.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-8.

10. A computer-readable storage medium for storing computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the method according to any one of claims 1-8.

Images