A histogram-based image noise estimation method and system
By using a histogram-based image noise estimation method, which utilizes local feature statistics and difference judgment, the problem of high computational cost and complexity in existing technologies is solved, achieving accurate and simple noise estimation applicable to images and videos of various formats.
Patent Information
- Application Number
- CN202311453461.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-11-03
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2043-11-03
AI Technical Summary
Existing image noise estimation techniques are computationally intensive and complex, making it difficult to effectively distinguish between the original image signal and noise.
A histogram-based image noise estimation method is adopted. By combining local feature statistics and difference judgment with window division of different sizes, the difference of local feature values is calculated, the noise level is estimated by using the law of large numbers, and the noise estimate is selected by smoothing filtering.
It achieves accurate noise estimation with low computational cost and simple calculation process. It can effectively distinguish noise from edge texture and meet the noise estimation requirements of Y, U, and V channel images or videos of different formats.
Smart Images

Figure CN117830129B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of image processing, and in particular to a histogram-based image noise estimation method and system. Background Art
[0002] In the process of image acquisition, the CMOS sensor will introduce various noises due to the influence of photodiodes, photon characteristics and circuit structure. The Raw format photo data collected by the sensor will be processed by many modules such as interpolation, gamma correction, color correction, white balance, local tone mapping, sharpening, etc., and the noise in the YUV domain is often large.
[0003] At this time, image denoising is needed. However, the difficulty of image denoising is how to distinguish the original image signal from the noise. In order to better distinguish the signal from the noise, it is necessary to first estimate the noise level of the current image so that there is a benchmark for the denoising strength. The existing technical solutions for image noise estimation are computationally intensive and the calculation process is relatively complicated. Summary of the invention
[0004] In order to solve the above technical problems, the present application provides a histogram-based image noise estimation method and system, which performs local feature statistics based on the histogram to achieve noise estimation in the YUV domain.
[0005] Specifically, the technical solution of this application is as follows:
[0006] In a first aspect, the present application discloses a histogram-based image noise estimation method, comprising the following steps:
[0007] Based on the original image to be processed, obtaining a Y channel image, a U channel image, and a V channel image of the original image;
[0008] Each pixel in the target channel image is processed in turn, and two windows are divided with the target pixel as the center to obtain the first window and the second window;
[0009] Calculating local features in the first window and the second window respectively to obtain a first eigenvalue and a second eigenvalue;
[0010] Calculate the difference between the first eigenvalue and the second eigenvalue; if the difference is less than a target threshold, determine that the local feature of the current window is related to noise, and perform statistics on the local feature; process each pixel point in the target channel image in turn, and obtain a histogram of the local feature by statistics;
[0011] The histogram is smoothed and filtered, and the maximum data value in the histogram is selected as the noise estimation value of the target channel image.
[0012] In some implementations, the first window and the second window have different window sizes.
[0013] In some implementations, the counting of the local features specifically includes the following steps:
[0014] Counting the first eigenvalues of the first window centered at the target pixel;
[0015] Or, statistics are collected on the second eigenvalues of the second window centered at the target pixel.
[0016] In some implementations, the local feature is the standard deviation of all pixel values in the current window; or the variance; or the sum of the absolute values of the differences.
[0017] In some implementations, the local feature of the current window is calculated using the following formula:
[0018] The variance calculation formula is as follows:
[0019] Where X is the image in the current window, M and N are the length and width of the window respectively; μ is the mean of all pixel values in the current window;
[0020] The standard deviation calculation formula is as follows:
[0021] Where X is the image in the current window, M and N are the length and width of the window respectively; μ is the mean of all pixel values in the current window;
[0022] The sum of the absolute values of the differences is calculated as follows:
[0023] Where X is the image in the current window, M and N are the length and width of the window respectively; μ is the mean of all pixel values in the current window;
[0024] The calculation formula of the mean is as follows:
[0025] Among them, X is the image in the current window, M and N are the length and width of the window respectively.
[0026] In some implementations, if the difference between the first eigenvalue and the second eigenvalue is greater than a target threshold, it is determined that the local feature of the current window is irrelevant to noise, and statistics are not performed on the local feature of the current window.
[0027] In a second aspect, the present application further discloses a histogram-based image noise estimation system, which is used to execute a histogram-based image noise estimation method described in any one of the above embodiments, comprising:
[0028] A preprocessing module, used for obtaining a Y channel image, a U channel image and a V channel image of the original image based on the original image to be processed;
[0029] A window division module is used to process each pixel point of the target channel image in turn, and perform two window divisions with the target pixel point as the center to obtain a first window and a second window;
[0030] A calculation module, used to calculate the local features in the first window and the second window respectively to obtain a first eigenvalue and a second eigenvalue;
[0031] The calculation module is further used to calculate the difference between the first eigenvalue and the second eigenvalue;
[0032] A statistical module is used for determining that the local feature of the current window is related to the noise if the difference is less than the target threshold, and performing statistics on the local feature; and after processing each pixel point in the target channel image in turn, obtaining a histogram of the local feature by statistics;
[0033] The filtering module is used to perform smoothing filtering on the histogram and select the maximum data value in the histogram as the noise estimation value of the target channel image.
[0034] In some implementations, the first window and the second window have different window sizes.
[0035] In some implementations, the statistical module is used to collect statistics on the first feature value of the first window centered on the target pixel point;
[0036] Alternatively, the statistical module is further used to collect statistics on the second eigenvalues of the second window centered on the target pixel point.
[0037] In some implementations, the local feature is the standard deviation of all pixel values in the current window; or the variance; or the sum of the absolute values of the differences.
[0038] Compared with the prior art, the present invention has at least one of the following beneficial effects:
[0039] 1. This application proposes to perform histogram statistics on the local eigenvalues of noise. Perform point-by-point operations on the image, calculate the difference between two windows of different sizes by considering the local characteristics of the noise, and then judge whether the conditions are met. This application fully considers the distribution characteristics of the noise. According to the law of large numbers, the average value of the sample will tend to the expected value of the real noise level, and then estimate the noise level. The method of this application can effectively distinguish noise from edge texture by judging the local feature difference. Accurate noise estimation can be achieved.
[0040] 2. This application uses the standard deviation, variance, and the sum of the absolute values of the differences of the pixel values in the corresponding window, or the multiple of the standard deviation, variance, and the sum of the absolute values of the differences, one of which is used as the basic data of the local feature. This application also includes smoothing the statistically obtained histogram and selecting the maximum data value in the histogram as the noise estimation value of the target channel image. The calculation amount is small and the calculation process is simple. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] The preferred implementation scheme will be described below in a clear and understandable manner with reference to the accompanying drawings to further illustrate the above-mentioned characteristics, technical features, advantages and implementation methods of the present application.
[0042] Figure 1 A flowchart of the steps of a method embodiment of the present application;
[0043] Figure 2 A flowchart of another method embodiment of the present application;
[0044] Figure 3 This is a structural block diagram of a system embodiment of the present application. DETAILED DESCRIPTION
[0045] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.
[0046] It should be understood that when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections.
[0047] In order to simplify the drawings, only the parts related to the application are schematically shown in each figure, and they do not represent the actual structure of the product. In addition, in order to simplify the drawings and facilitate understanding, in some figures, only one of the parts with the same structure or function is schematically drawn or marked. In this article, "one" not only means "only one", but also means "more than one".
[0048] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.
[0049] In this document, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.
[0050] In addition, in the description of the present application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.
[0051] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the specific implementation methods of the present application will be described below with reference to the accompanying drawings. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings and other implementation methods can be obtained based on these drawings without creative work.
[0052] Images and videos go through various processing processes during their entire life cycle from acquisition to playback, such as acquisition, editing, encoding, transcoding, transmission, and display. Each process will introduce distortion. "Noise" is a common distortion introduced during signal acquisition. Image noise reduction technology reduces or removes the noise generated by the image to make the image more subjective.
[0053] The role of image noise estimation is to estimate the noise level in an image, which is very important for tasks such as image denoising and image enhancement. By estimating the noise level in an image, we can choose appropriate denoising algorithms and parameters to achieve better denoising effects. At the same time, image noise estimation can also be used for image quality assessment to help us determine whether the image is clear, sharp, and realistic.
[0054] Reference Manual Attached Figure 1 An embodiment of a histogram-based image noise estimation method provided by the present application comprises the following steps:
[0055] S100, based on an original image or video frame to be processed, obtaining a brightness map and a chromaticity map of the original image, wherein the brightness map includes a Y channel image, and the chromaticity map includes a U channel image and a V channel image;
[0056] S200, process each pixel in the target channel image in turn, divide the window twice with the target pixel as the center, and obtain the first window and the second window; specifically, any one of the three channel images Y, U, and V can be selected as the target channel image. Perform point-by-point operation on the target channel image, and the pixel being processed is the target pixel.
[0057] S300, respectively calculating local features in the first window and the second window to obtain a first eigenvalue and a second eigenvalue;
[0058] S400 specifically includes: S410, calculating the difference between the first eigenvalue and the second eigenvalue; S420, if the difference is less than the target threshold, determining that the local feature of the current window is related to noise, and performing statistics on the local feature; S430, if the difference between the first eigenvalue and the second eigenvalue is greater than the target threshold, determining that the local feature of the current window is not related to noise, and not performing statistics on the local feature of the current window. S440, after processing each pixel point in the target channel image in turn, statistically obtain a histogram of the local feature;
[0059] Specifically, the present application fully considers the distribution characteristics of noise. According to the law of large numbers, the average value of the sample will tend to the expected value of the real noise level, and then the noise level is estimated. Most of the pixels of an image (except for the overexposed area) contain noise, and the noise level is not much different. The signal strength of different pixels may vary greatly. Some pixels may be edges, some may be textures, and some may be flat areas, but they all contain almost the same noise level. Local features such as standard deviation are often closely related to noise. The greater the noise, the greater the local difference value, but the texture edge will also cause the local difference value to become larger, so we use two different windows. Different windows often have larger signal differences, but the noise level does not differ much. If the local feature difference between two different windows exceeds a certain threshold range, it is not considered to be caused by noise, but may be caused by the signal, so it is not allowed to participate in the histogram statistics of the noise.
[0060] S500, performing smoothing filtering on the histogram, and selecting the maximum data value in the histogram as the noise estimation value of the target channel image.
[0061] Specifically, in the process of histogram statistics, the statistical values are organized into a series of well-defined bars or bins, whose values are feature statistics calculated from the data, which may be brightness, gradient or any other features.
[0062] Smoothing the histogram is a relatively mature existing technology. Based on spatial domain filtering technology, its purpose is to average the grayscale values of neighboring pixels. The size of the neighborhood is directly related to the smoothing effect. The larger the neighborhood, the better the smoothing effect. However, if the neighborhood is too large, smoothing will cause greater loss of edge information, making the output image blurred. Generally, simple averaging or Gaussian smoothing can be used; the smoothed histogram is mapped to the original image to obtain a smoothed image. It should be noted that histogram smoothing filtering will cause the loss of image detail information, so in practical applications, the filtering parameters need to be selected according to the specific situation.
[0063] More preferably, this application can meet the noise estimation of Y, U, V channel images or videos in various formats such as yuv444, yuv422, yuv420, etc. After obtaining the noise estimation result of a single channel, the noise estimation results of two or three channels can be combined according to the design requirements to meet the noise estimation of different requirements.
[0064] In the first implementation of the present embodiment, based on the above-mentioned embodiment, the window sizes of the first window and the second window are different. The length and width of the first window are greater than the length and width of the second window. In the above-mentioned step S400, the statistics of the local features specifically include: statistics of the first eigenvalue of the first window centered on the target pixel point; since the size of the first window is larger, the accuracy of statistics using the first eigenvalue calculated in the first window is also higher. Calculation with the help of a larger window is the best implementation method. In the second implementation of the present embodiment, statistics of the second eigenvalue of the second window centered on the target pixel point are performed. The same technical effect as the first implementation of the present embodiment can also be achieved, but the accuracy is slightly inferior.
[0065] Another embodiment of a histogram-based image noise estimation method provided by the present application is shown in the attached specification. Figure 2 As shown, the following steps are included:
[0066] 1. First obtain the brightness map (Y channel) and chromaticity map (U and V channels) of the input image / video.
[0067] 2. Calculate the local eigenvalues of the 5x5 window and 3x3 window of the brightness image Y, chromaticity image U, and chromaticity image V respectively; obtain the first eigenvalue noiseV0 and the second eigenvalue noiseV1. For details, refer to Figure 2 , taking the Y channel image (brightness map Y) as an example. After InputY, the 5x5 window local difference calculation and the 3x3 window local difference calculation are performed on the brightness map Y respectively, and then the threshold is judged.
[0068] 3. Calculate the absolute value of the difference between NoiseV0 and NoiseV1. When it is less than the threshold Th, NoiseV0 is included in the histogram statistics. When it is greater than or equal to the threshold Th, NoiseV0 is not included in the histogram statistics.
[0069] 4. Finally, perform one-dimensional filtering on the histogram, smooth it, and then find the global maximum value of the histogram, which is used as the estimated value of the noise. For details, refer to Figure 2 After the threshold is judged, the histogram is counted, the histogram is filtered, and the global maximum value is found to obtain the noise estimation value output.
[0070] Specifically, in this embodiment, since the first window is a 5x5 window and the second window is a 3x3 window, the first eigenvalue is counted as basic data of the histogram.
[0071] The local feature (first feature value) is the standard deviation of all pixel values in the current window (first window); or the variance; or the sum of the absolute values of the differences.
[0072] In some embodiments, the local feature is a multiple of the standard deviation of all pixel values in the current window; or a multiple of the variance; or a multiple of the sum of the absolute values of the differences. Taking the standard deviation STD as an example, the first eigenvalue NoiseV0 is equal to four times the standard deviation (STD), and then rounded. Specifically: NoiseV0 = round (std * 4). Similarly, the variance (variance) or the sum of the absolute values of the differences (SAD) can also be used as the basic data of the local feature.
[0073] In one implementation of this embodiment, in the Y channel image, the pixel value is the gray value of the pixel. The local feature is the standard deviation of the gray values of all pixels in the current window; or the variance; or the sum of the absolute values of the differences;
[0074] In another implementation of this embodiment, in the U and V channel images (chromaticity diagram U and chromaticity diagram V), the pixel value is the chromaticity value of the pixel point. The local feature is the standard deviation, variance, or the sum of the absolute values of the chromaticity values of all pixels in the current window.
[0075] In another embodiment of a histogram-based image noise estimation method of the present application, based on any of the above embodiments, the local feature is the standard deviation, variance, or the sum of the absolute values of all pixel values in the current window. The local feature of the current window is calculated by the following formula:
[0076] The variance calculation formula is as follows:
[0077] The standard deviation calculation formula is as follows:
[0078] The sum of the absolute values of the differences is calculated as follows:
[0079] Among them, μ is the mean of all pixel values in the current window; X is the image in the current window; M and N are the length and width of the window respectively.
[0080] Specifically, the calculation formula of the mean is as follows:
[0081] Preferably, in another implementation of this embodiment, X(i, j) represents the image in the current window. The local features of the current window are calculated by the following formula:
[0082] The variance calculation formula is as follows:
[0083] The standard deviation calculation formula is as follows:
[0084] The sum of the absolute values of the differences is calculated as follows:
[0085] Where μ is the mean of all pixel values in the current window, X is the image in the current window, M and N are the length and width of the window respectively;
[0086] The calculation formula of the mean is as follows:
[0087] Based on the same technical concept, the present application also discloses a histogram-based image noise estimation system, which can be used to implement any of the above-mentioned histogram-based image noise estimation methods. Specifically, an embodiment of the histogram-based image noise estimation system of the present application is as shown in the attached specification. Figure 3 As shown, including:
[0088] The preprocessing module 10 is used to obtain a Y channel image, a U channel image and a V channel image of the original image based on the original image to be processed;
[0089] The window division module 20 is used to process each pixel point of the target channel image in sequence, and perform two window divisions with the target pixel point as the center to obtain a first window and a second window;
[0090] A calculation module 30, used to calculate the local features in the first window and the second window respectively to obtain a first eigenvalue and a second eigenvalue;
[0091] The calculation module 30 is further used to calculate the difference between the first eigenvalue and the second eigenvalue;
[0092] A statistical module 40 is used for determining that the local feature of the current window is related to noise if the difference is less than a target threshold, and performing statistics on the local feature; and after processing each pixel point in the target channel image in turn, obtaining a histogram of the local feature by statistics;
[0093] The filtering module 50 is used to perform smoothing filtering on the histogram and select the maximum data value in the histogram as the noise estimation value of the target channel image.
[0094] In one implementation of this embodiment, the first window and the second window have different window sizes. The statistical module is used to collect statistics on the first feature value of the first window centered on the target pixel point;
[0095] In another implementation of this embodiment, the size of the first window is smaller than the size of the second window. The statistical module is further configured to collect statistics on the second characteristic value of the second window centered at the target pixel.
[0096] Specifically, both implementations can achieve similar technical effects, but using a larger window as the sampling window can make the calculation result more accurate.
[0097] The present application provides another embodiment of a histogram-based image noise estimation system. Based on the above system embodiment, the local feature is the standard deviation of all pixel values in the current window; or the variance; or the sum of the absolute values of the differences.
[0098] Specifically, the local feature of the current window is calculated by the following formula:
[0099] The variance calculation formula is as follows:
[0100] The standard deviation calculation formula is as follows:
[0101] The sum of the absolute values of the differences is calculated as follows:
[0102] Among them, μ is the mean of all pixel values in the current window; X is the image in the current window; M and N are the length and width of the window respectively.
[0103] Specifically, the calculation formula of the mean is as follows:
[0104] In the Y channel image, the pixel value is the gray value of the pixel. The local feature is the standard deviation, variance, or the sum of the absolute values of the gray values of all pixels in the current window.
[0105] In another implementation of this embodiment, in the U and V channel images (chromaticity diagram U and chromaticity diagram V), the pixel value is the chromaticity value of the pixel point. The local feature is the standard deviation, variance, or the sum of the absolute values of the chromaticity values of all pixels in the current window.
[0106] After calculating the first eigenvalue and the second eigenvalue, the calculation module 30 is also used to calculate the difference between the first eigenvalue and the second eigenvalue; if the difference is less than the target threshold, the local characteristics of the current window centered on the target pixel point on the surface are related to the noise and meet the statistical requirements.
[0107] In the target channel image, the image is operated point by point until all the pixels are processed. The local features of the window corresponding to all the pixels that meet the statistical requirements are counted and summarized to obtain a histogram. The histogram is smoothed and filtered, and the maximum data value in the histogram is selected as the noise estimation value of the target channel image.
[0108] The image noise estimation system based on histogram described in this embodiment can obtain the noise estimation result of the target channel image by performing the above calculation steps on the target channel image. After obtaining the noise estimation result of a single channel, the noise estimation results of two or three channels can be combined according to design requirements to meet noise estimation of different requirements.
[0109] The histogram-based image noise estimation method and system of the present application have the same technical concept, and the technical details of the embodiments of the two are applicable to each other. In order to reduce repetition, they will not be repeated here.
[0110] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned program modules is used as an example for illustration. In actual applications, the above-mentioned functions can be assigned to different program modules as needed, that is, the internal structure of the device can be divided into different program units or modules to complete all or part of the functions described above. The program modules in the embodiment can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into a processing unit, and the above-mentioned integrated unit can be implemented in the form of hardware or in the form of software program units. In addition, the specific names of the program modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application.
[0111] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0112] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed with hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0113] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. Exemplarily, the device embodiments described above are merely schematic. Exemplarily, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation. Exemplarily, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the device or unit can be electrical, mechanical or other forms.
[0114] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0115] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0116] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.
[0117] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.
Claims
1. A histogram-based image noise estimation method, characterized in that: The steps include: Based on the original image to be processed, obtaining a Y channel image, a U channel image, and a V channel image of the original image; Each pixel in the target channel image is processed in turn, and two windows are divided with the target pixel as the center to obtain the first window and the second window; Calculating local features in the first window and the second window respectively to obtain a first eigenvalue and a second eigenvalue; the local feature is the standard deviation, variance, or the sum of absolute values of all pixel values in the current window; Calculate the difference between the first eigenvalue and the second eigenvalue; if the difference is less than a target threshold, determine that the local feature of the current window is related to noise, and perform statistics on the local feature; If the difference between the first eigenvalue and the second eigenvalue is greater than the target threshold, it is determined that the local feature of the current window is irrelevant to the noise, and the local feature of the current window is not counted; After processing each pixel point in the target channel image in turn, a histogram of local features is obtained by statistics; The histogram is smoothed and filtered, and the maximum data value in the histogram is selected as the noise estimation value of the target channel image.
2. The method for estimating image noise based on histogram according to claim 1, characterized in that: The first window and the second window have different window sizes.
3. The method for estimating image noise based on histogram according to claim 1 or 2, characterized in that: The said performing statistics on the said local features specifically comprises the following steps: Counting the first eigenvalues of the first window centered at the target pixel; Or, statistics are collected on the second eigenvalues of the second window centered at the target pixel.
4. The method for estimating image noise based on histogram according to claim 1, characterized in that: The local feature of the current window is calculated by the following formula: The variance calculation formula is as follows: Where X is the image in the current window, M and N are the length and width of the window respectively; μ is the mean of all pixel values in the current window; The standard deviation calculation formula is as follows: Where X is the image in the current window, M and N are the length and width of the window respectively; μ is the mean of all pixel values in the current window; The sum of the absolute values of the differences is calculated as follows: Where X is the image in the current window, M and N are the length and width of the window respectively; μ is the mean of all pixel values in the current window; The calculation formula of the mean is as follows: Among them, X is the image in the current window, M and N are the length and width of the window respectively.
5. A histogram-based image noise estimation system, characterized in that: Used to perform the histogram-based image noise estimation method according to any one of claims 1 to 4, comprising: A preprocessing module, used for obtaining a Y channel image, a U channel image and a V channel image of the original image based on the original image to be processed; A window division module is used to process each pixel point of the target channel image in turn, and perform two window divisions with the target pixel point as the center to obtain a first window and a second window; A calculation module, used to calculate the local features in the first window and the second window respectively to obtain a first eigenvalue and a second eigenvalue; the local feature is the standard deviation, variance, or the sum of the absolute values of all pixel values in the current window; The calculation module is further used to calculate the difference between the first eigenvalue and the second eigenvalue; A statistical module is used for determining that the local feature of the current window is related to noise if the difference is less than a target threshold, and performing statistics on the local feature; if the difference between the first eigenvalue and the second eigenvalue is greater than the target threshold, determining that the local feature of the current window is not related to noise, and not performing statistics on the local feature of the current window; and after processing each pixel point in the target channel image in turn, obtaining a histogram of the local feature by statistics; The filtering module is used to perform smoothing filtering on the histogram and select the maximum data value in the histogram as the noise estimation value of the target channel image.
6. The histogram-based image noise estimation system according to claim 5, characterized in that: The first window and the second window have different window sizes.
7. The histogram-based image noise estimation system according to claim 6, characterized in that: The statistical module is used to collect statistics of the first characteristic value of the first window centered at the target pixel point; Alternatively, the statistical module is further used to collect statistics on the second eigenvalues of the second window centered on the target pixel point.
Citation Information
Patent Citations
Method and device for detecting noise level of image
CN105809665A
Image noise intensity estimation method for Gaussian noise
CN107230208A
Image noise intensity estimation method, storage medium and processing and recognition device
CN109584204A
Method and system for removing salt and pepper noise of Bayer image
CN116503293A
Image noise intensity estimation method, image noise intensity estimation device, and image recognition device
US20200118248A1
