An image denoising method based on selective smoothing filtering

Abstract

The present application aims to provide a kind of image denoising method based on selective smoothing filtering, comprising the following steps: A, input the gray image of the image to be denoised, establish 3*3 rectangular frame, with rectangular frame moving sweeps the entire image to be denoised, calculate the gradient t of each pixel point in the center of rectangular frame and each pixel point on the edge of rectangular frame after each movement;Set noise threshold x, if |t|≤x, then the rectangular frame is input to step B;Otherwise, the rectangular frame is input to step C;B, set edge detection threshold n, calculate threshold K based on each rectangular frame input in step A;If K≥n, then the root mean square value h of gray value of the rectangular frame is calculated E , replace the gray value of the center point E of the rectangular frame;If K

Description

TECHNICAL FIELD

[0001] The present application relates to the field of image processing, in particular to the application in different view angle images, and specifically to an image denoising method based on selective smoothing filtering. BACKGROUND

[0002] The image of a digital signal will be disturbed by the noise of imaging equipment and external environment in the process of digital processing and transmission. The size of the noise is a very important factor for measuring the quality of the image, so measuring the noise and filtering out the noise quickly and accurately without affecting the overall performance of the system is an important method to improve the quality of the image. The noise of the image can be divided into system noise and environmental noise in terms of source. The system noise is the noise from the system itself, and the environmental noise is the noise from the external environment. In order to obtain an image of high quality, the image usually needs to be processed and optimized by an image denoising algorithm.

[0003] An important difficulty in image denoising is the control of the smoothing degree. If the smoothing degree is too low, the denoising will not be sufficient, and if the smoothing degree is too high, the details of the image will be lost, which is not natural in vision. Since the balance between insufficient denoising and excessive smoothing is difficult to grasp, and the judgment of the pros and cons of denoising varies from person to person, most denoising filters often allow users to adjust parameters to selectively correct the denoising result. Therefore, how to control the smoothing degree is an important problem in selective image denoising.

[0004] In related technologies, when the image is denoised, the same filter is usually used for all images, that is, the part of the image without noise will also be processed, resulting in that the image without noise is also disturbed by the filter, affecting the image quality. In addition, when the image contains multiple noises, since the multiple noises are not selectively processed, the processing of one kind of noise in the image will also interfere with the processing of other noises, for example, when the system noise is processed, the processing of the environmental noise will be affected.

[0005] Therefore, how to selectively process the noise contained in the image, select the corresponding fast and accurate smoothing filter for different noises, and improve the efficiency of image denoising processing have become technical problems to be solved. SUMMARY

[0006] The present application aims to provide an image denoising method based on selective smoothing filtering, which can selectively select the best area to process the image noise, avoid using the filter algorithm to filter the entire image, and cause the phenomenon of image blur. At the same time, the selective smoothing filtering can accurately control the smoothing filtering degree, effectively improve the denoising effect, and ensure the quality of the image.

[0007] The technical scheme of the present application is as follows:

[0008] The image noise reduction method based on selective smoothing filtering comprises the following steps:

[0009] A. input the image to be denoised, convert it into a gray image, and establish a 3*3 rectangular frame as follows:

[0010]

[0011] Starting from any position in the image to be denoised, the rectangular frame moves across the entire image to be denoised, and the distance of each movement is one pixel position. After each movement, the gray values of the pixel points in the rectangular frame are recorded. The gray value of the pixel point at the center of the rectangular frame is subtracted from the gray values of the pixel points at the edges of the rectangular frame, respectively, to obtain the gradient t of each edge pixel point of the rectangular frame at this position. Set a noise threshold x. If |t|≤x, it means that all the pixel points in the rectangular frame are normal points, and the pixel points in the rectangular frame are input to step B. Otherwise, it means that there are noise points in the pixel points in the rectangular frame, and the pixel points in the rectangular frame are input to step C.

[0012] B. Set an edge detection threshold n, and input the gray values of the pixel points of each rectangular frame input from step A into the following formula to calculate the threshold K.

[0013] (1)

[0014] If K≥n, the gray values of the edge pixel points of the rectangular frame are substituted into the following formula (2) to obtain the root mean square value h E , which replaces the gray value of the center point E of the rectangular frame. If K E , the pixel points of the rectangular frame are input to step C.

[0015] (2)

[0016] C. Perform Gaussian smoothing processing on each rectangular frame input from step A or step B.

[0017] D. After steps B and C are completed, the final denoised image is obtained.

[0018] The setting of the threshold x comprises the following steps:

[0019] Based on the rectangular frame established in step A, starting from any place in the image to be denoised, the rectangular frame moves across the entire image to be denoised, and each time the moving distance is one pixel position, and the gray value of each pixel in the rectangular frame is recorded after each movement, each rectangular frame is subjected to Gaussian smoothing, and the gray value of the center pixel of each Gaussian-smoothed rectangular frame is subtracted from the gray value of each pixel on the edge of the rectangular frame to obtain the gradient t0 of each edge pixel in each rectangular frame, and the maximum value of |t0| is selected, and the maximum value is set as the threshold value x.

[0020] The formula of the Gaussian smoothing is:

[0021] (3)

[0022] Wherein, sigma represents the weight of unit distance; x, y represent the template coordinates of the center pixel E.

[0023] The setting of the threshold value n includes the following steps:

[0024] The average value P of the gray value of all pixels of the original image is calculated, and then the threshold value x is brought into the following formula (4) to solve, and the calculation result is the threshold value n;

[0025] (4)

[0026] Wherein, X represents the center pixel value in the search frame; x is the set noise threshold value.

[0027] The present application carries out rectangular frame sweeping segmentation on the image to be denoised, carries out specific selection on the segmented pixel region, and carries out specific filtering processing, so as to filter different types of noise, avoid image blur caused by filtering all images, and highlight the edge by root mean square denoising and smoothing of the edge contour, so as to ensure the denoising effect and the image quality. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 The original picture before denoising of the plastic part;

[0029] Figure 2 The picture after denoising of the plastic part. DETAILED DESCRIPTION

[0030] The present application will be specifically described below in combination with the drawings and examples. Example 1

[0031] The image denoising method based on selective smoothing filtering includes the following steps:

[0032] A, input the image to be denoised, convert it into a gray-scale image, and establish a 3*3 rectangular frame as follows:

[0033]

[0034] From any starting point in the image to be denoised, the rectangular frame moves across the entire image to be denoised, and each time the moving distance is one pixel position. After each movement, the gray scale values of each pixel in the rectangular frame are recorded. The gray scale values of the center pixel of the rectangular frame and the gray scale values of each pixel on the edge of the rectangular frame are respectively subtracted to obtain the gradient t of each edge pixel in the rectangular frame at this position;

[0035] The noise threshold x is set, and the specific process is as follows: based on the rectangular frame established in step A, starting from any point in the image to be denoised, the rectangular frame moves across the entire image to be denoised, and each time the moving distance is one pixel position. After each movement, the gray scale values of each pixel in the rectangular frame are recorded. Each rectangular frame is subjected to Gaussian smoothing, and the gray scale values of the center pixel of each Gaussian-smoothed rectangular frame and the gray scale values of each pixel on the edge of the rectangular frame are respectively subtracted to obtain the gradient t0 of each edge pixel in each rectangular frame. The maximum value of |t0| is selected, and the maximum value is set as the threshold x.

[0036] For each rectangular frame, if |t|≤x, it represents that the pixels in the range of the rectangular frame are all normal points, and the pixels in the rectangular frame are input to step B; otherwise, it represents that there are noise points in the pixels in the range of the rectangular frame, and the pixels in the rectangular frame are input to step D;

[0037] B, set the edge detection threshold n, and input the gray scale values of each pixel of each rectangular frame input in step A into the following formula to calculate the threshold K;

[0038] (1)

[0039] The setting of the threshold n includes the following steps:

[0040] The average value P of the gray scale values of all pixels of the original image is calculated, and then the average value P and the threshold x are brought into the following formula (4) to solve, and the calculation result is the threshold n;

[0041] (4)

[0042] Wherein, X represents the center pixel value in the search frame; x is the set noise threshold;

[0043] If K≥n, the gray scale values of each edge pixel of the rectangular frame are brought into the following formula (2) to calculate the root mean square value h E , and the root mean square value h E is substituted for the gray scale value of the center point E of the rectangular frame; if K

[0044] (2);

[0045] C, Gaussian smoothing is performed on each rectangular frame input in step A or step B;

[0046] The formula of Gaussian smoothing is:

[0047] (3)

[0048] Wherein, σ represents the weight of unit distance; x, y represent the template coordinates of the center pixel point E.

[0049] D, after step B and C are calculated, the final denoising image is obtained.

[0050] The setting of the threshold value n includes the following steps:

[0051] The average value P of the gray value of all pixel points of the original image is calculated, and then the threshold value x is brought into the following formula (4) to solve the threshold value n.

[0052] Example 2

[0053] The plastic part picture is denoised by the method of example 1, and the specific results are shown in Figure 1 and Figure 2 , Figure 1 is the original picture before denoising of the part, Figure 2 is the denoised picture.

[0054] From Figure 1 and Figure 2 it can be seen that Figure 1 After denoising, the noise points basically disappear, and the image is clearer.

Claims

1. A method for image denoising based on selective smoothing filtering, characterized in that, It comprises the following steps: A. input the image to be denoised, convert it into a gray image, and establish a 3*3 rectangular frame as follows: ； Starting from any position in the image to be denoised, the rectangular frame moves across the entire image to be denoised, and each time the moving distance is one pixel position. After each movement, the gray values of the pixel points in the rectangular frame are recorded. The gray values of the pixel points on the edges of the rectangular frame are subtracted from the gray value of the pixel point at the center of the rectangular frame to obtain the gradient t of each edge pixel point in the rectangular frame. Set a noise threshold x. If |t|≤x, it means that the pixel points in the rectangular frame are normal points, and the pixel points in the rectangular frame are input to step B. Otherwise, it means that there are noise points in the pixel points in the rectangular frame, and the pixel points in the rectangular frame are input to step C. B. Set an edge detection threshold n, and input the gray values of the pixel points in each rectangular frame input from step A into the following formula to calculate the threshold K. （1） If K≥n, the gray value of each edge pixel of the rectangular frame is brought into the following formula (2) to obtain the root mean square value h E The root mean square value h E is substituted into the gray value of the center point E of the rectangular frame; if K （2）； C. Perform Gaussian smoothing on each rectangular frame input from step A or step B. D. After steps B and C are completed, the final denoised image is obtained.

2. The image denoising method based on selective smoothing filtering according to claim 1, wherein, The setting of the threshold x comprises the following steps: Based on the rectangular frame established in step A, starting from any position in the image to be denoised, the rectangular frame moves across the entire image to be denoised, and each time the moving distance is one pixel position. After each movement, the gray values of the pixel points in the rectangular frame are recorded. The gray values of the pixel points on the edges of the rectangular frame are subtracted from the gray value of the pixel point at the center of the rectangular frame to obtain the gradient t of each edge pixel point in the rectangular frame. Set a noise threshold x. If |t|≤x, it means that the pixel points in the rectangular frame are normal points, and the pixel points in the rectangular frame are input to step B. Otherwise, it means that there are noise points in the pixel points in the rectangular frame, and the pixel points in the rectangular frame are input to step C.

3. The image denoising method based on selective smoothing filtering according to claim 1 or 2, characterized in that: The formula of the Gaussian smoothing is: （3） Where σ represents the weight of the unit distance; x and y represent the template coordinates of the center pixel point E.

4. The image denoising method based on selective smoothing filtering according to claim 2, wherein, The setting of the threshold n comprises the following steps: The average value of the gray values of all pixel points of the original image is calculated as P, which is then input into the following formula (4) together with the threshold x to solve the threshold n. （4） Where X represents the center pixel value in the search frame; x is the set noise threshold.

Images