CN116703746A

CN116703746A - A filtering method and related device for removing ultra-high-density salt-and-pepper noise polluted images

Info

Publication number: CN116703746A
Application number: CN202310428500.4A
Authority: CN
Inventors: 周军妮; 董颖娣; 李昌华; 江莉; 王燕妮; 杨润玲; 毛力; 史丽晨; 石昊琦
Original assignee: Xian University of Architecture and Technology
Current assignee: Xian University of Architecture and Technology
Priority date: 2023-04-20
Filing date: 2023-04-20
Publication date: 2023-09-05

Abstract

A filtering method and a related device for removing ultra-high density salt and pepper noise pollution images comprise the following steps: acquiring a noise image, and performing edge mirror image expansion on the noise image to obtain an edge expansion noise image; performing noise detection on the edge-spread noise image, and recording the result in a noise identification matrix; acquiring a filtering template, selecting a corresponding filtering template aiming at noise points, and combining an edge expanded noise image to obtain a filtering function; and determining a filtered image when the noise recognition matrix is judged to be 0 based on the filtering function. Firstly, a strategy of detecting and filtering is adopted, so that signal points are not polluted; and secondly, a cross-like filtering template is adopted, so that the neighborhood signal points of the to-be-filtered point are fully utilized, and the detail information of the filtered image can be recovered to a great extent.

Description

Filtering method for removing ultra-high density spiced salt noise pollution image and related device

Technical Field

The invention belongs to the technical field of signal processing, and particularly relates to a filtering method and a related device for removing ultra-high density salt and pepper noise pollution images.

Background

Images are often contaminated with impulse noise due to noise and errors generated in the image sensor or communication channel transmission. Therefore, it is important to remove image noise before performing subsequent processing such as edge detection, image segmentation, and object recognition.

The traditional method for removing impulse noise is to apply a median filter, which is a nonlinear signal processing technology based on order statistics theory, and can filter impulse noise while maintaining the image edge. However, the conventional median filter cannot distinguish noise from signal points, which is hardly effective for high-density noise images. Thus, over the past twenty years, numerous scholars have conducted extensive research into median-based filters to remove image impulse noise. In addition, in order to avoid signal points from being contaminated, recently, many documents apply a switching scheme in which noise point detection is performed first, and then only the detected noise points are filtered to avoid signal points from being contaminated.

A progressive switching median filter (Progressive Switching Median Filter: PSMF) is one of the earliest filters that applied the switching strategy based removal of impulse noise. However, this approach uses iterative thresholds for noise detection, resulting in missed detection and increased time costs. To solve the problem of leakage detection in PSMF, eng and Ma have proposed noise adaptive soft-switching median filters (Noise Adaptive Soft-Switching Median Filter: NASMF) using the concept of fuzzy logic. The scheme realizes the intellectualization of noise detection and reduces the omission ratio. However, as Noise Density (ND) becomes greater, selecting a larger filter window can result in severe image blurring and be time-costly. For a G-level gray-level noise image, the corrupted pixels are typically digitized into two extrema: minimum or maximum in dynamic range (0 or G-1). For this reason, impulse noise usually appears as white or black spots in the image, which is the source of salt-and-pepper noise. Based on this, many filters determine the maximum and minimum of the image dynamic range as noise points, such as a simple adaptive median filter (Simple Adaptive Median Filter: SAMF), a fast high efficiency median filter (Fast and Efficient Median Filter: FEMF), a clipped median filter based on pixel density (Pixel Density Based Trimmed Median Filter: PDBTMF), and an adaptive neighborhood extreme processing mean filter (Adaptive Neighborhood Winsorized Mean Filter: ANWMF). For most natural images, this noise detection method is simple and effective. But not all pixels with intensity 0 or G-1 are noise. In view of this, a modified clipping median filter based on probability decisions (Probabilistic Decision-Based Improved Trimmed Median Filter: PDITMF) and a pixel density based filter (Based on Pixel Density Filter: BPDF), which correspond to a point with a value of 0 or G-1, make a conditional determination as to whether the point is a noise point.

SAMF adaptively selects a filter window according to ND, and has simple algorithm principle, easy implementation, but too severe conditions. As ND increases, a larger filtering template is required. At this time, the correlation of the filter points with the neighborhood decreases, resulting in the filtered image becoming more blurred. The same problem exists with FEMF and BPDF. In order to solve the problem of excessive filter templates and loss of more detail, both PDBTMF and ANWMF employ a fixed 3 filter window. When ND >90%, the filter window may be a full noise window, the PDBTMF algorithm is completely disabled, and ANWMF directly assigns a certain gray value to the condition, so that the filtered restored image has a plurality of color blocks with the same gray value, and image details are lost.

Disclosure of Invention

The invention aims to provide a filtering method and a related device for removing an ultra-high density salt and pepper noise pollution image, which are used for solving the problems that when the existing salt and pepper noise image filtering algorithm is used for low density salt and pepper noise, the filtering effect is still good, and once the noise density exceeds 90%, the filtering algorithm is almost invalid.

In order to achieve the above purpose, the present invention adopts the following technical scheme:

in a first aspect, the present invention provides a filtering method for removing ultra-high density pretzel noise pollution images, including:

acquiring a noise image, and performing edge mirror image expansion on the noise image to obtain an edge expansion noise image;

performing noise detection on the edge-spread noise image, and recording the result in a noise identification matrix;

acquiring a filtering template, selecting a corresponding filtering template aiming at noise points, and combining an edge expanded noise image to obtain a filtering function;

and determining a filtered image when the noise recognition matrix is judged to be 0 based on the filtering function.

Optionally, edge mirror expansion is performed on the noise image:

the design edge mirror image expansion function is shown as (1):

wherein y' represents an edge mirror extension image of the noise image y, and the size is (l+6) × (w+6); y' _kg The gray values at the points (k, g) k, g ε B, B= { (k, g) |1. Ltoreq.k. Ltoreq.L+6, 1. Ltoreq.g. Ltoreq.W+6 }; the area to be treated is (k, g) k, g e B ', B' = { (k, g) |4. Ltoreq.k. Ltoreq.L+3, 4. Ltoreq.g. Ltoreq.W+3 }.

Optionally, noise detection is performed on the edge-spread noise image:

for each (k, g) in the region B', a noise detection matrix M is designed in order from left to right and from top to bottom ^NI As shown in formula (2):

wherein ,the numbers of points with gray scale of 0 and 255 in w×w window are represented by (k, g) as the center, and the value of w is 7.

Optionally, a filtering template is obtained:

let R1-R6 represent euclidean distance from the point to be filtered, and specifically take the value r1=1;R3＝2；R6＝3；R _ED representing the Euclidean distance of the point (k, g) to be filtered from the neighborhood (m, n); omega shape ₁ ～Ω ₅ Respectively representing cross-like filtering templates used by the algorithm of the invention;

optionally, trimming the template definition;

and y' (m, n))≠0),y'(m,n)≠255)},k＝1,2,3,4,5

Trimming the definition of the number of template elements;

the function card representsThe number of elements in the series, the result is recorded in the variable +.>

Optionally, for the noise point, selecting a corresponding filtering template, and combining the edge-expanded noise image to obtain a filtering function:

the filter function is expressed as:

optionally, when the noise recognition matrix is determined to be 0 based on the filtering function, determining a filtered image:

within region B', each is determined following a left-to-right, top-to-bottom principleDetermining a current round of filtered image f;

taking f as the new noise image y, i.e. performing y=f, if corresponding to M ^NI Repeatedly executing the whole flow if the matrix is not zero; if corresponding to M ^NI And (5) a zero matrix, and ending the filtering.

In a second aspect, the present invention provides a filtering system for removing ultra-high density pretzel noise contaminated images, comprising:

the mirror image expansion module is used for acquiring a noise image, carrying out edge mirror image expansion on the noise image, and obtaining an edge expansion noise image;

the noise detection module is used for carrying out noise detection on the edge expansion noise image, and the result is recorded in the noise identification matrix;

the filtering function acquisition module is used for acquiring a filtering template, selecting a corresponding filtering template aiming at noise points, and combining the edge-expanded noise image to obtain a filtering function;

and the image determining module is used for determining a filtered image when the noise identification matrix is judged to be 0 based on the filtering function.

In a third aspect, the present invention provides a computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of a filtering method for removing ultra-high density salt and pepper noise contaminated images when executing the computer program.

In a fourth aspect, the present invention provides a computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of a filtering method for removing ultra-high density pretzel noise contaminated images.

Compared with the prior art, the invention has the following technical effects:

firstly, a strategy of detecting and filtering is adopted, so that signal points are not polluted; secondly, a cross-like filtering template is adopted, so that neighborhood signal points of the to-be-filtered points are fully utilized, and detail information of the filtered image can be recovered to a great extent; finally, the filtering scheme provided by the invention is convenient for hardware implementation. The filtering effect is superior to the existing similar filtering method from subjective evaluation to objective evaluation.

Drawings

FIG. 1 is a cross-like filtering template used in the algorithm of the present invention, shown in FIG. (a) Ω ₁ ；(b)Ω ₂ ；(c)Ω ₃ ；(d)Ω ₄ ；(e)Ω ₅ ；

Fig. 2 adds ND to the test image Lena: at 90%, the filtering results of the traditional Median Filtering (MF) and the Haidi adaptive median filtering algorithm (SAMF) and the adaptive switching median filtering Algorithm (ASMMBF) provided by the invention are compared with each other (figure (a) Lena original figure; b) noise image (ND: 90%); (c) MF, (d) SAMF; e) ASMMBF)

Fig. 3 adds ND to the test image Lena: 95% of the results of the MF, SAMF and ASMMBF filtering are compared with the figure (a) Lena original figure, (b) noise image (ND: 95%); (c) MF, (d) SAMF and (e) ASMMBF

Fig. 4 adds ND to the test image Lena: 99% of the comparison image of the MF, SAMF and ASMMBF filtering results (figure (a) Lena original image, (b) noise image (ND: 99%); (c) MF, (d) SAMF and (e) ASMMBF)

FIG. 5 is a test image Hand X-ray film Hand X-ray Add ND:90% of the comparison image of the MF, SAMF and ASMMBF filtering results (a) Hand X-ray original image, (b) noise image (ND: 95%); (c) MF, (d) SAMF and (e) ASMMBF

FIG. 6 is a Hand X-ray film Hand X-ray add ND:95% of the comparison image of the MF, SAMF and ASMMBF filtering results (a Hand X-ray original image, (b) noise image (ND: 95%); (c) MF, (d) SAMF and (e) ASMMBF

FIG. 7 is a test image Hand X-ray film Hand X-ray Add ND:99% of the comparison image of the MF, SAMF and ASMMBF filtering results (a Hand X-ray original image, (b) noise image (ND: 99%); (c) MF, (d) SAMF and (e) ASMMBF)

Fig. 8 is a test image sailing boat Sailboats add ND:90% of the results of the MF, SAMF and ASMMBF filtering are compared with the images (a) Sailbots original image, (b) noise image (ND: 95%); (c) MF, (d) SAMF and (e) ASMMBF

Fig. 9 is a test image sailing boat Sailboats with ND added: 95% of the results of the MF, SAMF and ASMMBF filtering are compared with the images (a) Sailbots original image, (b) noise image (ND: 95%); (c) MF, (d) SAMF and (e) ASMMBF

Fig. 10 is a test image sailing boat Sailboats with ND added: 99% of the comparison image of the MF, SAMF and ASMMBF filtering results (a Sailbots original image, (b) noise image (ND: 99%); (c) MF, (d) SAMF and (e) ASMMBF

Fig. 11 is a test image Fingerprint finger print add ND:90% of the result of the MF, SAMF and ASMMBF filtering is compared with the figure (a) of the primary Fingerprint, (b) of the noise image (ND: 95%); (c) of the MF, (d) of the SAMF and (e) of the ASMMBF

Fig. 12 adds ND to the test image Fingerprint finger print: 95% of the results of the MF, SAMF and ASMMBF filtering are compared with the images (a) Sailbots original image, (b) noise image (ND: 95%); (c) MF, (d) SAMF and (e) ASMMBF

Fig. 13 is a test image Fingerprint finger print add ND:99%, comparing the MF, SAMF and ASMMBF filtering results with a figure (a) a Fingerprint original figure, (b) a noise image (ND: 99%); (c) MF, (d) SAMF and (e) ASMMBF;

fig. 14 is an artificial test image nine palace lattice added ND:99%, comparing the MF, SAMF and ASMMBF filtering results with the figure (a) Sudoku original figure, (b) noise image (ND: 99%), (c) MF, (d) SAMF and (e) ASMMBF;

FIG. 15 is a graph comparing PSNR of test images;

fig. 16 is a comparison chart of the test image SSIM.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and examples. The method specifically comprises the following steps:

and step one, carrying out edge mirror image expansion on the noise image y. The design edge mirror image expansion function is shown as (1):

y' represents an edge mirror extension image of the noise image y, and has a size of (l+6) × (w+6). y' _kg Represents the gray value at the point (k, g) (k, g ε B, B= { (k, g) |1. Ltoreq.k. Ltoreq.L+6, 1. Ltoreq.g. Ltoreq.W+6). The area to be treated is (k, g) (k, g E B ', B' = { (k, g) |4. Ltoreq.k. Ltoreq.L+3, 4. Ltoreq.g. Ltoreq.W+3 });

and step two, carrying out noise detection on the edge expansion noise image y'. For each (k, g) in the region B', a noise detection matrix M is designed in order from left to right and from top to bottom ^NI As shown in formula (2):

Step three, designing a filtering template;

(1) Defining a filtering template;

let r1=1;R3＝2；R6＝3；R _ED representing the euclidean distance of the point to be filtered (k, g) from the neighborhood (m, n). Omega shape ₁ ～Ω ₅ Respectively representing cross-like filtering templates used by the inventive algorithm.

(2) Trimming template definition;

and y '(m, n) noteq0), y' (m, n) noteq255), k=1, 2,3,4,5 (4)

(3) Trimming the definition of the number of template elements;

Step four, selecting a corresponding filtering template aiming at noise points, and combining the steps shown in the figure (1) to obtain a filtering functionThe design is as follows:

step five, determining a current round of filtering image;

within region B', each is determined according to equation (6) following the left-to-right, top-to-bottom principleDetermining a current round of filtered image f according to formula (7);

taking f as the new noise image y, i.e. performing y=f, if corresponding to M ^NI Repeatedly executing the first to fifth steps if the matrix is not zero; if corresponding to M ^NI If the matrix is zero, the filtering is finished;

the following are specific examples given by the inventors;

the traditional median filtering and other numerous improved algorithms based on median filtering remove salt and pepper noise in the image, and the denoising effect is good under the condition of low noise density, however, as the noise density ND is increased, particularly when the noise density is higher than 90%, the filtering algorithm is completely invalid. The embodiment provides a filtering algorithm for removing high-density salt and pepper noise pollution images. This process is an adaptive cyclic operation process. The method comprises the steps of firstly carrying out edge mirror image expansion on a noise image, and aiming at processing image edge pixels during filtering, then applying the adaptive switching median mean value filtering algorithm ASMMBF to carry out filtering, and finally taking the filtering result of the round as a new noise image to continuously judge whether the next round of filtering is carried out. Simulation experiments show that compared with the traditional median filtering method, the method has higher peak signal-to-noise ratio (Peak Signal to Noise Ratio:PSNR) and structural similarity index (Structural Similarity:SSIM), and meanwhile subjective evaluation and objective evaluation indexes PSNR and SSIM result are assisted, so that the method is convenient for research and application in subsequent image analysis, image identification and detection.

The test images are selected from digital images with different types and different resolutions, including portrait, X-ray film, landscape, fingerprint and artificial image for subjective evaluation of algorithm.

Fig. 2,3 and 4 are graphs comparing filtering effects when 90%, 95% and 99% of salt and pepper noise are added to the test portrait image Lena. Wherein, (a) resolution is 512 to 512 clear Lena image, (b) noise image, (c) traditional median filtering MF effect diagram, (d) self-adaptive median filtering SAMF effect diagram, and (e) self-adaptive switching median average filtering ASMMBF effect diagram. The graph shows that after high-density noise (ND: 90%, 95%, 99%) is added, the noise completely floods the signal itself, and the MF algorithm fails to filter when ND is 90%, 95%, 99%. The SAMF algorithm can identify the portrait when ND is 90%, however, the whole image has black spots, and the detail edge of the portrait is blurred; when ND is 95%, the outline of the portrait is thin and visible, and the whole image is covered by a large black spot; when ND is 99%, the outline of the portrait is thin and visible, and the whole image is covered by a large black spot. The ASMMB algorithm provided by the invention has obviously better definition than MF and SAMF, and restores the detail characteristics of the image while filtering, and particularly, the recognizable portrait image is still restored when ND is 99%.

Fig. 5, 6 and 7 are graphs comparing the filtering effect of the hand X-ray film test with 90%, 95% and 99% salt and pepper noise added respectively. Wherein, (a) resolution is 420 to 344 clear hand X-ray film, (b) noise image, (c) traditional median filtering MF filtering effect image, (d) an adaptive median filter (SAMF) filtering effect diagram, and (e) an algorithm of the invention adaptively switches the ASMMBF filtering effect diagram. The graph shows that after high-density noise (ND: 90%, 95%, 99%) is added, the noise completely floods the signal itself, and the image target is completely unrecognizable. The MF algorithm fails in average filtering at 90%, 95%, 99% ND. The SAMF algorithm can identify the shape of the hand bone when ND is 90%, however, the edges of the finger bone and the metacarpal bone are not smooth, obvious saw teeth are formed, the edges of the wrist bone are completely blurred, the hand bone cannot be distinguished, and a small amount of black plaques exist in the filter map; when ND is 95%, the outline of the hand is invisible, and the whole image is covered by a large black spot; at 99% ND, the filter is almost covered by dark color, with a small number of bright spots in the hand area. The ASMMB algorithm provided by the invention has obviously better definition than MF and SAMF, and restores the detail characteristics of the image while filtering, and particularly, the distinguishable hand image is still restored when ND is 99%.

Fig. 8, 9 and 10 are graphs comparing the filtering effects of the test sailing boat landscape when the ND is 90%, 95% and 99% salt-pepper noise is added respectively. The resolution ratio of (a) is 768 to 512 clear sailboat landscape images, (b) noise images, (c) traditional median filtering MF effect images, (d) self-adaptive median filtering SAMF effect images, and (e) the algorithm self-adaptive switching median average filtering ASMMBF effect images. The graph shows that after high-density noise (ND: 90%, 95%, 99%) is added, the noise completely floods the signal itself, and the MF algorithm fails to filter when ND is 90%, 95%, 99%. SAMF algorithm can be recognized by sailing boat when ND is 90%, however, the whole image has a small amount of black spots, and the detailed edge of sailing boat is blurred; when ND is 95%, the contour of the sailing boat is thin and visible, and the whole image is covered by a large black spot; at 99% ND, the image is covered entirely by a dark color, the sailing boat outline is completely invisible, and the algorithm fails. The ASMMB algorithm filtering effect is obviously superior to MF and SAMF, the detail characteristics of the image are recovered while filtering, when ND is 90%, the ship body, the mast, the sail and the boat doctor are all clearly visible in the figure, and especially when ND is 99%, the distinguishable sailing figure is recovered.

Fig. 11, 12 and 13 are graphs comparing filtering effects when 90%, 95% and 99% of salt and pepper noise is added to the test fingerprint graph. Wherein, (a) resolution is 480 multiplied by 400 clear images, (b) noise images, (c) traditional median filtering MF effect images, (d) self-adaptive median filtering SAMF effect images, and (e) algorithm self-adaptive switching median average filtering ASMMBF effect images. The graph shows that after high-density noise (ND: 90%, 95%, 99%) is added, the noise completely floods the signal itself, and the MF algorithm fails to filter when ND is 90%, 95%, 99%. The SAMF algorithm can identify fingerprints when ND is 90%, however, the whole image has black spots, and the fingerprint texture is intermittent and incomplete; when ND is 95%, the fingerprint contour is thin and visible, and the whole image is covered by a large black spot; at 99% ND, the fingerprint is completely invisible and the image is covered in its entirety by a dark color. The algorithm ASMMB of the invention has obviously better definition than MF and SAMF, restores the detail characteristics of the image while filtering, and particularly still restores the distinguishable fingerprint image when ND is 99%.

Fig. 14 (a) shows an artificial nine-square image with a resolution of 600×600, 9 squares with a resolution of 200×200 are visible, and the gray values are 30, 200, 120, respectively, corresponding to the three colors black, white, and gray in the figure. (b) Nine Gong Getu for adding 99% of salt-pepper noise, illustrating that the image noise completely covers the signal, (c) illustrates that the algorithm is completely invalid due to the traditional median filtering effect, (d) illustrates that the SAMF filtering effect is achieved, and a small number of bright spots exist in the bright grid, so that the nine-grid pattern is unrecognizable; (e) For the ASMMB filtering effect of the algorithm, the nine-grid pattern is obvious, and the three-color black, white and gray scale is displayed normally.

The objective evaluation adopts peak signal-to-noise ratio PSNR and structural similarity index SSIM for evaluation, and is specifically defined as follows:

in the formulas (8) and (9), x represents an original clear image, f represents a filtering result obtained by adding salt and pepper noise to x, and L, W respectively represents a filtering recovery image; mu (mu) _x 、μ _f Represents the average value of x and f, delta _x 、δ _f Representing the variance of x, f, delta _xf Covariance of x, f; for g=256, c ₁ The value is 6.5025, C ₂ The value is 58.5225.

Table 1 shows PSNR and SSIM values of different algorithms when 90%, 95% and 99% salt-pepper noise were added to 4 natural test images and 1 artificial image, respectively.

Fig. 15 is a graph comparing PSNR and SSIM values of the test image under 3 ND conditions, and the graph shows that the algorithm of the present invention has a very high signal-to-noise ratio at a high noise density.

Fig. 16 is a comparison graph of PSNR and SSIM values for a test image at 3 ND, showing that the filtered image and the original sharp image have high structural similarity at high noise density for the algorithm of the present invention.

In an embodiment of the present invention, a filtering system for removing an ultra-high density salt and pepper noise pollution image is provided, which can be used to implement the filtering method for removing an ultra-high density salt and pepper noise pollution image, and specifically, the system includes:

The division of the modules in the embodiments of the present invention is schematically only one logic function division, and there may be another division manner in actual implementation, and in addition, each functional module in each embodiment of the present invention may be integrated in one processor, or may exist separately and physically, or two or more modules may be integrated in one module. The integrated modules may be implemented in hardware or in software functional modules.

In yet another embodiment of the present invention, a computer device is provided that includes a processor and a memory for storing a computer program including program instructions, the processor for executing the program instructions stored by the computer storage medium. The processor may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf Programmable gate arrays (FPGAs) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., which are the computational core and control core of the terminal adapted to implement one or more instructions, in particular adapted to load and execute one or more instructions within a computer storage medium to implement the corresponding method flow or corresponding functions; the processor provided by the embodiment of the invention can be used for the operation of a filtering method for removing ultra-high density salt and pepper noise pollution images.

In yet another embodiment of the present invention, a storage medium, specifically a computer readable storage medium (Memory), is a Memory device in a computer device, for storing a program and data. It is understood that the computer readable storage medium herein may include both built-in storage media in a computer device and extended storage media supported by the computer device. The computer-readable storage medium provides a storage space storing an operating system of the terminal. Also stored in the memory space are one or more instructions, which may be one or more computer programs (including program code), adapted to be loaded and executed by the processor. The computer readable storage medium herein may be a high-speed RAM memory or a non-volatile memory (non-volatile memory), such as at least one magnetic disk memory. One or more instructions stored in a computer-readable storage medium may be loaded and executed by a processor to implement the corresponding steps of the above-described embodiments with respect to a filtering method for removing ultra-high density pretzel noise contaminated images.

It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

Finally, it should be noted that: the above embodiments are only for illustrating the technical aspects of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the above embodiments, it should be understood by those of ordinary skill in the art that: modifications and equivalents may be made to the specific embodiments of the invention without departing from the spirit and scope of the invention, which is intended to be covered by the claims.

Claims

1. A filtering method for removing ultra-high density salt and pepper noise pollution images is characterized by comprising the following steps:

2. The filtering method for removing ultra-high density pretzel noise contaminated image according to claim 1, wherein edge mirror expansion is performed on the noisy image:

the design edge mirror image expansion function is shown as (1):

3. The filtering method for removing ultra-high density impulse noise contaminated image according to claim 2, wherein the noise detection is performed on the edge-spread noise image:

4. The filtering method for removing ultra-high density salt and pepper noise pollution images as claimed in claim 1, wherein a filtering template is obtained:

let R1-R6 represent euclidean distance from the point to be filtered, and specifically take the value r1=1;R3＝2；R6＝3；R _ED representing the Euclidean distance of the point (k, g) to be filtered from the neighborhood (m, n); omega shape ₁ ～Ω ₅ Respectively represent the cross-like shapes used by the inventive algorithmFiltering templates;

5. the filtering method for removing ultra-high density pretzel noise contaminated images according to claim 4, wherein the template definition is pruned;

trimming the definition of the number of template elements;

6. The filtering method for removing ultra-high density salt and pepper noise pollution images according to claim 1, wherein for noise points, a corresponding filtering template is selected, and the noise images are expanded by combining edges to obtain a filtering function:

7. the filtering method for removing ultra-high density impulse noise contaminated image according to claim 6, wherein when the noise recognition matrix is determined to be 0 based on the filtering function, the filtered image is determined:

8. A filtering system for removing ultra-high density pretzel noise contaminated images, comprising:

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the computer program, carries out the steps of a filtering method for removing ultra high density salt and pepper noise contaminated images according to any one of the claims 1 to 7.

10. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the steps of a filtering method for removing ultra high density salt and pepper noise contaminated images according to any one of the claims 1 to 7.