Image processing method and device, and storage medium
By performing multiple filtering processing and texture intensity calculation on the Y, U and V components of the image, an adaptive image denoising method is realized, which solves the problem of difficult to balance texture retention and noise removal in the prior art, and improves the image quality after denoising.
Patent Information
- Application Number
- CN202510167100.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-14
- Publication Date
- 2025-06-03
AI Technical Summary
The existing image denoising methods are difficult to adaptively process local textures of images, making it difficult to achieve a balance between retaining textures and removing noise, affecting the image quality after denoising.
By acquiring the Y, U and V components of the image, the first filtering process is performed several times to obtain multiple images of smoothness, the texture intensity corresponding to each smoothness is calculated, and the second filtering process is performed on the components of each smoothness is performed. Then, the image after the second filtering process is fused with the image of the corresponding smoothness according to the texture intensity to obtain the denoised target image.
The local texture of the image is realized adaptively processed, and the denoising intensity of the local area of the image is adaptively determined, and the balance between retaining the texture and removing noise is achieved, ensuring the quality of the target image after denoising.
Smart Images
Figure CN120088161A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of image denoising, and particularly to an image processing method, an apparatus, and a storage medium. Background Art
[0002] Image noise is a random signal generated by various interferences during image acquisition, transmission, or processing, manifested as random fluctuations in pixel values in the image. These fluctuations generally reduce the quality of the image, affecting the visual effect of the image and subsequent processing. Therefore, image denoising solutions have always been a research hotspot in the field of image processing. Current denoising methods generally operate directly in the pixel space of the image, that is, the so-called spatial domain denoising methods, and the filtering methods used are mean filtering, median filtering, Gaussian filtering, bilateral filtering, Wiener filtering, wavelet filtering, non-local mean filtering, etc. Although these filtering methods have their respective advantages, there are also some inherent disadvantages, that is, they cannot adaptively process the local texture of the image, that is, they cannot adaptively determine the denoising intensity of the local area of the image, so it is difficult to achieve a balance between retaining the texture and removing the noise, resulting in poor quality of the target image after denoising. Summary of the Invention
[0003] In view of this, the present application provides an image processing method, an apparatus, and a storage medium, which can improve the problem that the existing image processing methods are difficult to adaptively process the local texture of the image and the resulting difficulty in achieving a balance between retaining the texture and removing the noise.
[0004] An image processing method provided by the present application includes:
[0005] Obtaining the Y, U, and V components of the original image;
[0006] Performing multiple first filtering processes on the Y, U, and V components respectively to obtain images with multiple smoothness levels correspondingly;
[0007] Calculating the texture intensity of the Y, U, and V components corresponding to each smoothness level according to a preset threshold;
[0008] Performing a second filtering process on the Y, U, and V components corresponding to each smoothness level respectively;
[0009] Fusing the image after the second filtering process with the image corresponding to the smoothness level according to the texture intensity to obtain a denoised target image.
[0010] Optionally, the first filtering process is a series of multi-channel non-directional filtering processes, and the second filtering process includes at least one non-directional filtering process.
[0011] Optionally, for any one of the Y, U, and V components, the window sizes and intensities of the multiple first filtering processes are different.
[0012] Optionally, the window sizes and intensities for performing the first filtering process on the Y, U, and V components are different.
[0013] Optionally, the number of cascaded filters for performing the first filtering process on the Y, U, and V components is different.
[0014] Optionally, the preset threshold includes a first threshold and a second threshold, and the first threshold is greater than the second threshold;
[0015] Calculating the texture intensities of the Y, U, and V components corresponding to each smoothness level according to the preset threshold includes:
[0016] Obtaining the absolute value of the component difference corresponding to each smoothness level;
[0017] If the absolute value is greater than or equal to the first threshold, determining that the texture intensity is a first value;
[0018] If the absolute value is less than or equal to the second threshold, determining that the texture intensity is a second value;
[0019] If the absolute value is less than the first threshold and greater than the second threshold, obtaining the ratio of the difference between the absolute value and the second threshold to the difference between the first threshold and the second threshold as the texture intensity.
[0020] Optionally, the component difference is the difference between the component corresponding to the previous smoothness level and the component corresponding to the current smoothness level, where for the first smoothness level, the component corresponding to the previous smoothness level is the component corresponding to the original image.
[0021] Optionally, the method further includes: performing a second filtering process on the Y, U, and V components of the original image;
[0022] Performing the second filtering process on the Y, U, and V components corresponding to each smoothness level respectively includes: performing the second filtering process on the Y, U, and V components corresponding to other smoothness levels except the last smoothness level.
[0023] Optionally, fusing the image after the second filtering process with the image corresponding to the corresponding smoothness level according to the texture intensity includes:
[0024] Obtaining a first product of the component after the second filtering process and the corresponding texture intensity, and a second product of the component after the first filtering process and the value obtained by subtracting the corresponding texture intensity from 1;
[0025] Taking the sum of the first product and the second product as the fusion intermediate quantity corresponding to each smoothness level;
[0026] Obtain the output components corresponding to the Y, U, and V components according to the fused intermediate quantity.
[0027] Optionally, the obtaining the output components corresponding to the Y, U, and V components according to the fused intermediate quantity includes:
[0028] S = (S n *(1 - t n-1 ) + S n-1 *t n-1 )*(1 - t n-2 ) + S n-2 *t n-2
[0029] where S is the output component, S n is the fused intermediate quantity corresponding to the last smoothness level, n is the serial number of the last smoothness level, t n-1 is the texture intensity of the component corresponding to the previous smoothness level, S n-1 is the fused intermediate quantity corresponding to the previous smoothness level, t n-2 is the texture intensity of the component corresponding to the second previous smoothness level, and S n-1 is the fused intermediate quantity corresponding to the second previous smoothness level.
[0030] An image processing device provided by the present application includes a processor and a memory. An image processing program is stored on the memory. When the image processing program is executed by the processor, the steps of the image processing method described in any one of the above are implemented.
[0031] A storage medium provided by the present application stores a computer program. When the computer program is executed by a processor, the steps of the image processing method described in any one of the above are implemented.
[0032] As described above, the present application calculates the texture intensities of the Y, U, and V components corresponding to each smoothness level, and further filters the Y, U, and V components corresponding to each smoothness level respectively (i.e., performs the second filtering process), and then fuses the image after the further filtering process with the image corresponding to the smoothness level according to the texture intensity to obtain the denoised target image. In this way, the present application can adaptively process the local texture of the image, that is, adaptively determine the denoising intensity of the local area of the image, achieve a balance between retaining the texture and removing the noise, and ensure the quality of the denoised target image. And, the texture intensity is calculated according to a preset threshold, so that the denoising intensity can be adjusted by adjusting the preset threshold. In addition, the texture intensity is obtained based on the result of the filtering process (i.e., the first filtering process), making full use of the existing information, so that the calculation time, the hardware implementation area, and the operating power consumption can be reduced. Description of the Drawings
[0033] Figure 1It is a schematic flowchart of an image processing method according to an embodiment of the present application;
[0034] Figure 2 It is a schematic flowchart of a method provided by the present application for performing a first filtering process and a second filtering process on the Y component. Detailed implementation manners
[0035] To solve the above problems existing in the prior art, the present application provides an image processing method, an apparatus, and a storage medium. These several protection subjects are based on the same concept, and the principles for solving problems are basically the same or similar. The implementation manners of each protection subject can be referred to each other, and the repeated parts will not be elaborated.
[0036] In the solutions of each protection subject of the present application, the texture intensities of the Y, U, and V components corresponding to each smoothness level are calculated, and the Y, U, and V components corresponding to each smoothness level are further filtered respectively. Then, according to the texture intensities, the image after the further filtering process is fused with the image corresponding to the smoothness level to obtain a denoised target image. Herein, the present application can adaptively process the local texture of the image, that is, adaptively determine the denoising intensity of the local area of the image, and achieve a balance between retaining the texture and removing the noise to ensure the quality of the denoised target image.
[0037] To make the objectives, technical solutions, and advantages of the present application clearer, the technical solutions of the present application will be clearly described below in conjunction with specific embodiments and the corresponding drawings. Obviously, the embodiments described below are only a part of the embodiments of the present application, rather than all the embodiments. Without conflict, the following various embodiments and their technical features can be combined with each other, and they also belong to the technical solutions of the present application.
[0038] Figure 1 It is a schematic flowchart of an image processing method according to an embodiment of the present application. This image processing method can also be referred to as a "method", and is at least applicable to denoising an image. The execution subject of each step can be an image processing device, such as a mobile intelligent terminal such as a mobile phone or a PAD, a non-mobile intelligent terminal such as a computer, or a storage medium, a processor, etc. with image processing functions.
[0039] As Figure 1 shown, the method at least includes the following steps S1 to S5.
[0040] S1: Obtain the Y, U, and V components of the original image.
[0041] The original image can be regarded as a to-be-processed image to be denoised. The Y, U, and V components respectively represent the luminance information and chrominance information of the original image. Among them, the Y component represents the luminance or brightness of the original image, and the U and V components represent the chrominance of the original image, which are used to describe the color and saturation of the image.
[0042] The Y, U, and V components can identify that the original image adopts the YUV color encoding format. If the original image is in the RGB color encoding format, it is necessary to first convert the original image from the RGB color encoding format to the YUV color encoding format. The specific conversion process and principle can refer to the prior art and will not be elaborated here. Optionally, the YUV color encoding format of the present application can be any one of YUV444, YUV422, and YUV420. In the description herein, unless otherwise specifically or clearly indicated, the YUV420 YUV color encoding format of the original image is taken as an example for illustration.
[0043] Denosing based on the YUV space can enable the present application to have at least the following advantages: First, the YUV space separates the luminance and chrominance information, allowing for targeted denoising processing of the luminance and chrominance signals, thereby improving the flexibility of denoising; Second, compared with the RGB space, the chrominance signal in the YUV space usually has a lower sampling rate, which reduces the computational complexity of the denoising algorithm; Third, by separately processing the luminance and chrominance signals, the edge and texture information of the image can be better protected; moreover, when denoising the chrominance signal, a milder filtering method can be adopted to reduce color distortion.
[0044] S2: Perform multiple first filtering processes on the Y, U, and V components respectively to correspondingly obtain images with multiple levels of smoothness.
[0045] The multiple levels of smoothness can be regarded as the smoothness at multiple stages, and the smoothness at each stage is different.
[0046] In one example, the first filtering process is a series of multi-channel non-directional filtering processes, and the non-directional filtering process includes, but is not limited to, mean filtering or Gaussian filtering. In any first filtering process, the filtering methods of the series of multi-channel non-directional filtering processes can be different.
[0047] When performing the first filtering process on these three components, a filter can be designed according to specific requirements to remove or enhance certain frequency components in the original image. When removing noise, the present application achieves the expected effect by separately filtering the Y, U, and V components. Specifically, filtering the Y component can process the luminance information of the image, making the luminance distribution of the image more uniform or highlighting certain luminance features. Separately filtering the U and V components can process the chrominance information of the image, making the colors of the image more vivid or removing certain unwanted color interferences.
[0048] Taking the example of performing three consecutive non-directional filtering operations on each component: After performing three non-directional filtering operations on the Y component, the Y components corresponding to three images with different smoothness levels are obtained, namely Y_NonDir_1, Y_NonDir_2, and Y_NonDir_3; after performing three non-directional filtering operations on the U component, the U components corresponding to three images with different smoothness levels are obtained, namely U_NonDir_1, U_NonDir_2, and U_NonDir_3; after performing three non-directional filtering operations on the V component, the V components corresponding to three images with different smoothness levels are obtained, namely V_NonDir_1, V_NonDir_2, and V_NonDir_3.
[0049] In one example, step S2 may satisfy at least one of the following:
[0050] Feature 1: For any one of the Y, U, and V components, the window sizes and intensities of multiple first filtering operations are different.
[0051] Feature 2: The window sizes and intensities of the first filtering operations on the Y, U, and V components are different.
[0052] Feature 3: The number of cascaded filters for the first filtering operations on the Y, U, and V components is different.
[0053] For Feature 1 and Feature 2, that is to say, the window sizes and intensities of the three cascaded non-directional filters for each component can be set to different values, and the window sizes and intensities of the non-directional filters between the components do not need to be consistent. Taking the Y component as an example, the non-directional filter corresponding to the first smoothness level of the Y component can use a window of size 7x7 and a Gaussian filter with sigma = 2, and the non-directional filters corresponding to the second and third smoothness levels can use a window of size 5x5 and a Gaussian filter with sigma = 1.8.
[0054] For Feature 3, that is to say, the Y, U, and V components can use different numbers of cascaded filters, and the subsequent steps S3 to S5 are determined adaptively according to the number of cascaded filters.
[0055] S3: Calculate the texture intensities of the Y, U, and V components corresponding to each smoothness level according to a preset threshold.
[0056] In one example, the preset threshold includes two thresholds, respectively called the first threshold and the second threshold, and the first threshold is greater than the second threshold. Herein, step S3 includes:
[0057] S31. Obtain the absolute values of the component differences corresponding to each smoothness level;
[0058] S32. If the absolute value is greater than or equal to the first threshold, determine the texture intensity as the first value;
[0059] S33. If the absolute value is less than or equal to the second threshold, determine the texture intensity as the second value;
[0060] S34. If the absolute value is less than the first threshold and greater than the second threshold, obtain the ratio of the difference between the absolute value and the second threshold to the difference between the first threshold and the second threshold as the texture intensity.
[0061] In the aforementioned step S31, the component difference is the difference between the component corresponding to the previous smoothness level and the component corresponding to the current smoothness level. Among them, for the first smoothness level, the component corresponding to the previous smoothness level is the component corresponding to the original image.
[0062] Still taking the aforementioned example of performing three consecutive non-directional filtering processes on each component, for the Y component, there are three stages of smoothness levels:
[0063] Combined Figure 2 As shown, the calculation formula for the texture intensity corresponding to the smoothness level in the first stage is:
[0064] str_tmp_1 = abs(Y - Y_NonDir_1)
[0065] if str_tmp_1 >= thr_1_high
[0066] text_str_1 = 1
[0067] if str_tmp_1 <= thr_1_low
[0068] text_str_1 = 0
[0069] if str_tmp_1 > thr_1_low and str_tmp_1 < thr_1_high
[0070] text_str_1 = (str_tmp_1 - thr_1_low) / (thr_1_high - thr_1_low)
[0071] Among them, str_tmp_1 is the intermediate texture intensity corresponding to the smoothness degree in the first stage, Y is the Y value corresponding to the original image, Y - Y_NonDir_1 is the component difference corresponding to the smoothness degree in the first stage, abs is the absolute value of Y - Y_NonDir_1, that is, the absolute value, thr_1_high is the first preset threshold corresponding to the first stage, thr_1_low is the second preset threshold corresponding to the first stage, and thr_1_high > thr_1_low, and text_str_1 is the texture intensity corresponding to the smoothness degree in the first stage.
[0072] Combined with Figure 2 As shown, the calculation relationship formula for the texture intensity corresponding to the smoothness degree in the second stage is:
[0073] str_tmp_2 = abs(Y_NonDir_1 - Y_NonDir_2)
[0074] if str_tmp_2 >= thr_2_high
[0075] text_str_2 = 1
[0076] if str_tmp_2 <= thr_2_low
[0077] text_str_2 = 0
[0078] if str_tmp_2 > thr_2_low and str_tmp_2 < thr_2_high
[0079] text_str_2 = (str_tmp_2 - thr_2_low) / (thr_2_high - thr_2_low)
[0080] Among them, str_tmp_2 is the intermediate texture intensity corresponding to the smoothness degree in the second stage, (Y - Y_NonDir_1 - Y - Y_NonDir_2) is the component difference corresponding to the smoothness degree in the second stage, thr_2_high is the first preset threshold corresponding to the second stage, thr_2_low is the second preset threshold corresponding to the second stage, and thr_2_high > thr_2_low, and text_str_2 is the texture intensity corresponding to the smoothness degree in the second stage.
[0081] Combined with Figure 2 As shown, the calculation relationship formula for the texture intensity corresponding to the smoothness degree in the third stage is:
[0082] str_tmp_3 = abs(Y_NonDir_2 - Y_NonDir_3)
[0083] if str_tmp_3 >= thr_3_high
[0084] text_str_3 = 1
[0085] if str_tmp_3 <= thr_3_low
[0086] text_str_3 = 0
[0087] if str_tmp_3 > thr_3_low and str_tmp_3 < thr_3_high
[0088] text_str_3 = (str_tmp_3 - thr_3_low) / (thr_3_high - thr_3_low)
[0089] Among them, str_tmp_3 is the intermediate texture intensity corresponding to the smoothness degree in the third stage, (Y - Y_NonDir_2 - Y - Y_NonDir_3) is the component difference corresponding to the smoothness degree in the third stage, thr_3_high is the first preset threshold corresponding to the third stage, thr_3_low is the second preset threshold corresponding to the third stage, and thr_3_high > thr_3_low, and text_str_3 is the texture intensity corresponding to the smoothness degree in the third stage.
[0090] For the U component and the V component, the texture intensity corresponding to each smoothness degree can refer to the calculation principle and process of the texture intensity of each stage of the Y component, which will not be elaborated here.
[0091] S4: Perform second filtering processing on the Y, U, and V components corresponding to each smoothness degree respectively.
[0092] In an example, the step S4 refers to performing second filtering processing on the Y, U, and V components corresponding to other smoothness degrees except the last smoothness degree; the method further includes: performing second filtering processing on the Y, U, and V components of the original image respectively.
[0093] That is to say, combined with Figure 2As shown, taking the example of the above-mentioned non-directional filtering process of concatenating each component three times, for the Y component, in step S4, the components that need to be subjected to the second filtering process are Y, Y_NonDir_1, and Y_NonDir_2, and the Y components corresponding to the smoothness degree in the third stage are not respectively subjected to the second filtering process. The second filtering process includes at least one non-directional filtering process. The specific filtering method can be at least one of wavelet filtering, bilateral filtering, non-local mean filtering, guided filtering, etc. The corresponding filtering principle can refer to the prior art. The components after respectively performing the second filtering process on Y, Y_NonDir_1, and Y_NonDir_2 are respectively labeled as Y_Dir_1, Y_Dir_2, and Y_Dir_3.
[0094] Similarly, for the U component, in step S4, the components that need to be subjected to the second filtering process are U, U_NonDir_1, and U_NonDir_2. The components after respectively performing the second filtering process on U, U_NonDir_1, and U_NonDir_2 are respectively labeled as U_Dir_1, U_Dir_2, and U_Dir_3; for the V component, in step S4, the components that need to be subjected to the second filtering process are V, V_NonDir_1, and V_NonDir_2. The components after respectively performing the second filtering process on V, V_NonDir_1, and V_NonDir_2 are respectively labeled as V_Dir_1, V_Dir_2, and V_Dir_3.
[0095] S5: Fuse the image after the second filtering process with the image corresponding to the corresponding smoothness degree according to the texture intensity to obtain the denoised target image.
[0096] In one example, the above fusion can be performed through the following steps S51 to S53:
[0097] S51. Obtain the first product of the component after the second filtering process and the corresponding texture intensity, and the second product of the component after the first filtering process and 1 minus the corresponding texture intensity value;
[0098] S52. Take the sum of the first product and the second product as the fusion intermediate quantity corresponding to each smoothness degree;
[0099] S53. Obtain the output components corresponding to the Y, U, and V components according to the fusion intermediate quantity.
[0100] Optionally, step S53 can obtain the corresponding output components through the following relational expression:
[0101] S = (S n *(1 - t n-1 ) + S n-1 *t n-1 )*(1 - t n-2) + S n-2 *t n-2
[0102] Among them, S is the output component, and S n is the fusion intermediate quantity corresponding to the last smoothness level. n is the serial number of the last smoothness level. For example, if there are three stages of smoothness levels, n takes the value of 3 according to the sequential sorting of the smoothness levels, the value of n - 1 is 2, and the value of n - 2 is 1. t n-1 is the texture intensity of the component corresponding to the previous smoothness level, and S n-1 is the fusion intermediate quantity corresponding to the previous smoothness level, and t n-2 is the texture intensity of the component corresponding to the second previous smoothness level, and S n-1 is the fusion intermediate quantity corresponding to the second previous smoothness level.
[0103] Still taking the above-mentioned non-directional filtering process of three series connections for each component as an example, for the Y component, the fusion intermediate quantities Y_tmp_3, Y_tmp_2, Y_tmp_1 corresponding to the three stages of smoothness levels are obtained as follows:
[0104] Y_tmp_3 = Y_Dir_3 * text_str_3 + Y_NonDir_3 * (1 - text_str_3)
[0105] Y_tmp_2 = Y_Dir_2 * text_str_2 + Y_NonDir_2 * (1 - text_str_2)
[0106] Y_tmp_1 = Y_Dir_1 * text_str_1 + Y_NonDir_1 * (1 - text_str_1)
[0107] Among them, Y_Dir_3 * text_str_3 is the first product corresponding to the third stage of smoothness level, and Y_NonDir_3 * (1 - text_str_3) is the second product corresponding to the third stage of smoothness level.
[0108] The final output component S is marked as Y_out, and the obtaining method is:
[0109] Y_out_2 = Y_tmp_3 * (1 - text_str_2) + Y_tmp_2 * textstr_2
[0110] Y_out_1 = Y_out_2 * (1 - text_str_1) + Y_tmp_1 * text_str_1
[0111] Y_out = Y_out_1
[0112] Among them, Y_out_2 is the aforementioned S n *(1 - t n-1 ) + S n-1 *t n-1 ), Y_tmp_1 * text_str_1 is the aforementioned (1 - t n-2 ) + S n-2 *t n-2 .
[0113] For the U and V components, the output components U_out and V_out corresponding to each smoothness level can refer to the calculation principle and process of the output components of the Y component in each stage, which will not be elaborated here.
[0114] Y_out, U_out, and V_out are the Y, U, and V components obtained after fusion respectively. Based on this, the denoised target image can be obtained.
[0115] Based on the above, the present application calculates the texture intensity of the Y, U, and V components corresponding to each smoothness level, and further filters the Y, U, and V components corresponding to each smoothness level (i.e., performs the second filtering process). Then, according to the texture intensity, the image after the further filtering process is fused with the image corresponding to the smoothness level to obtain the denoised target image. Herein, the present application can adaptively process the local texture of the image, that is, adaptively determine the denoising intensity of the local area of the image, and achieve a balance between retaining the texture and removing the noise, ensuring the quality of the denoised target image. Moreover, the texture intensity is calculated according to a preset threshold. Herein, the denoising intensity can be adjusted by adjusting the preset threshold. In addition, the texture intensity is obtained based on the result of the filtering process (i.e., the first filtering process), making full use of the existing information. Herein, the calculation time, the hardware implementation area, and the operating power consumption can be reduced.
[0116] The embodiment of the present application also provides a storage medium, on which an image processing program is stored. This image processing program is essentially a computer program, and when this image processing program is executed by a processor, it realizes the steps of the image processing method in any example.
[0117] This storage medium includes but is not limited to any one of read-only memory (ROM, Read Only Memory), random access memory (RAM, Random Access Memory), magnetic disks, and optical discs.
[0118] Since the program stored in this storage medium can execute the steps in the image processing method provided by any embodiment of the present application, the beneficial effects that can be achieved by the image processing method in any of the foregoing embodiments can be realized. For details, see the foregoing embodiments, which will not be elaborated here.
[0119] The embodiments of the present application further provide an image processing device or chip, including a memory and a processor. An image processing program is stored on the memory. When the image processing program is executed by the processor, the steps of the image processing method in any of the foregoing embodiments are implemented; and / or, the image processing device or chip is provided with a storage medium as exemplified above, and the processor loads the storage medium to execute the steps of the image processing method, thereby achieving the beneficial effects that can be achieved by the image processing method in the corresponding embodiment.
[0120] The above are only some embodiments of the present application, and thus do not limit the patent scope of the present application. For those of ordinary skill in the art, any equivalent structural transformation made by using the content of this specification and the drawings shall similarly be included in the patent protection scope of the present application.
[0121] In this article, step codes such as S1 and S2 are used. The purpose is to more clearly and briefly express the corresponding content, and it does not constitute a substantial limitation in order. Those skilled in the art may execute S31 first and then S1 during specific implementation, etc., but these should all be within the protection scope of the present application.
[0122] Although the terms "first", "second", etc. are used in this article to describe various information, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. In addition, the singular forms "a", "an", and "the" are also intended to include the plural forms. The terms "or" and "and / or" are interpreted as inclusive, or mean any one or any combination. An exception to this definition only occurs when the combination of elements, functions, steps, or operations is inherently mutually exclusive in some way.
Claims
1. An image processing method, characterized in that: include: Get the Y, U and V components of the original image; Performing a plurality of first filtering processes on the Y, U and V components respectively, so as to obtain images with a plurality of smoothness levels; Calculate the texture intensity of the Y, U and V components corresponding to each smoothness level according to the preset threshold; Performing a second filtering process on the Y, U and V components corresponding to each smoothing degree respectively; The image after the second filtering process is fused with an image of a corresponding smoothness level according to the texture intensity to obtain a denoised target image.
2. The method according to claim 1, characterized in that The first filtering process is a series of multiple non-directional filtering processes, and the second filtering process includes at least one non-directional filtering process.
3. The method according to claim 1 or 2, characterized in that: Satisfy at least one of the following: For any one of the Y, U and V components, the window size and strength of the multiple first filtering processes are different; The window sizes and intensities of the first filtering process performed on the Y, U and V components are different; The number of series filters for performing the first filtering process on the Y, U and V components is different.
4. The method according to claim 1, characterized in that The preset threshold includes a first threshold and a second threshold, and the first threshold is greater than the second threshold; The calculating the texture strength of the Y, U and V components corresponding to each smoothness level according to the preset threshold comprises: Get the absolute value of the component difference corresponding to each smoothing degree; If the absolute value is greater than or equal to the first threshold, determining the texture intensity to be a first value; If the absolute value is less than or equal to the second threshold, determining the texture intensity to be a second value; If the absolute value is smaller than the first threshold and larger than the second threshold, a ratio of a difference between the absolute value and the second threshold to a difference between the first threshold and the second threshold is obtained as the texture intensity.
5. The method according to claim 4, characterized in that The component difference is the difference between the component corresponding to the previous smoothness level and the component corresponding to the current smoothness level, wherein, for the first smoothness level, the component corresponding to the previous smoothness level is the component corresponding to the original image.
6. The method according to claim 1, characterized in that The method further comprises: performing second filtering processing on the Y, U and V components of the original image respectively; The performing the second filtering process on the Y, U and V components corresponding to each smoothness level respectively includes: performing the second filtering process on the Y, U and V components corresponding to other smoothness levels except the last smoothness level respectively.
7. The method according to claim 6, characterized in that The method further comprises: fusing the image processed by the second filtering with an image of a corresponding smoothness according to the texture strength, comprising: Obtaining a first product of the second filtered component and a corresponding texture intensity, and a second product of the first filtered component and 1 minus a corresponding texture intensity value; The sum of the first product and the second product is used as the fusion intermediate amount corresponding to each smoothing degree; The output components corresponding to the Y, U and V components are obtained according to the fused intermediate quantity.
8. The method according to claim 7, characterized in that The step of obtaining output components corresponding to the Y, U and V components according to the fused intermediate quantity includes: S=(S n *(1-t n-1 )+S n-1 *t n-1 )*(1-t n-2 )+S n-2 *t n-2 Among them, S is the output component, S n is the fusion intermediate quantity corresponding to the last smoothing level, n is the sequence number of the last smoothing level, t n-1 is the texture intensity of the component corresponding to the previous smoothness level, S n-1 is the fusion intermediate quantity corresponding to the previous smoothness level, t n-2 is the texture intensity of the component corresponding to the second smoothness level, S n-1 It is the fusion intermediate amount corresponding to the second smoothness level.
9. An image processing device, characterized in that: The method comprises a processor and a memory, wherein an image processing program is stored in the memory, and when the image processing program is executed by the processor, the steps of the image processing method according to any one of claims 1 to 8 are implemented.
10. A storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a processor, the steps of the image processing method according to any one of claims 1 to 8 are implemented.