Method and apparatus for denoising image data

Abstract

The application discloses a denoising method and device of image data, relates to the technical field of image processing, and mainly aims to solve the problem of poor effectiveness of existing image denoising. The method comprises the following steps: acquiring standard image data collected under different environments; performing geometric correction on the standard image data to obtain corrected image data, and performing regional division based on the corrected image data to obtain a bright region image and a dark region image; constructing a denoising model based on the bright region image and the dark region image, and when target image data to be denoised is acquired, performing denoising on the target image data based on the denoising model to obtain denoised image data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to an image data denoising method and device. BACKGROUND

[0002] In the field of image processing, especially in high-precision imaging tasks such as hyperspectral imaging, remote sensing imaging, medical imaging, etc., image striping noise is a common and significant image quality problem. Among them, the striping noise usually presents as brightness or grayscale fluctuation along a fixed direction (such as horizontal or vertical direction), which is expressed as periodic or non-periodic light and dark banding. The causes mainly include the following aspects: detector nonuniformity: when the sensitivity or gain response of some detector units of the image sensor exists deviation, the grayscale values between different scanning lines will appear systematic difference, forming striping noise; electronic noise and time-varying interference: including current drift, voltage disturbance, etc. caused by temperature fluctuation, unstable power supply or electromagnetic interference in analog circuit; damage in compression and transmission process: some images may also cause grid structure disturbance due to data error or loss in compression or transmission stage, indirectly causing striping effect; illumination or incident angle change: under non-ideal shooting conditions, poor illumination uniformity may also exacerbate the expression of striping characteristics.

[0003] At present, the existing denoising methods can usually adopt calibration method, mean elimination method or image restoration method. However, the calibration method depends on the internal parameters of the equipment and has strong scene dependence, which is difficult to cope with time-varying disturbance or complex background change in the actual imaging process; the mean elimination method is prone to over-smoothing, edge blurring and even image distortion under the conditions of obvious coexistence of bright and dark areas in the image, complex structure edges, etc., and cannot finely preserve the image structure information; the image restoration method usually needs a large amount of high-quality labeled data for training, and the generalization ability is limited by the representativeness of the training set, and the denoising effect is poor. SUMMARY

[0004] Therefore, the present application provides an image data denoising method and device, which mainly aims to solve the problem of poor effectiveness of the existing image denoising.

[0005] According to one aspect of the present application, an image data denoising method is provided, comprising:

[0006] acquiring standard image data collected under different environments;

[0007] performing geometric correction on the standard image data to obtain corrected image data, and performing regional division based on the corrected image data to obtain bright area images and dark area images;

[0008] construct a denoising model based on the bright area image and the dark area image, and when target image data to be denoised is acquired, denoising the target image data based on the denoising model to obtain denoised image data.

[0009] Further, the geometric correction of the standard image data to obtain corrected image data comprises:

[0010] stripes of the standard image data are converted by a row-by-row correction method to obtain standard image data of a regular strip structure;

[0011] on the condition that a first row of pixels of the standard image data of the regular strip structure is determined as a reference pixel, the mean square error of the first row of pixels and each row of pixels is calculated by a sliding window method;

[0012] the standard image data of the regular strip structure is corrected according to the offset pixel corresponding to the minimum mean square error to obtain the corrected image data.

[0013] Further, before the geometric correction of the standard image data to obtain corrected image data, the method further comprises:

[0014] the standard image data is subjected to random noise elimination by the minimum pixel value of each row of the standard image data, so as to be the standard image data to be geometrically corrected.

[0015] Further, the region division based on the corrected image data to obtain a bright area image and a dark area image comprises:

[0016] the region division of the corrected image data is performed by at least one of a histogram peak value, a local gray fluctuation, a pixel density distribution feature, and a global adaptive threshold algorithm to obtain a bright area image and a dark area image.

[0017] Further, before the denoising model is constructed based on the bright area image and the dark area image, the method further comprises:

[0018] the average value of the region pixels of the bright area image and the dark area image is calculated respectively to obtain a bright area average pixel and a dark area average pixel.

[0019] Further, before the denoising model is constructed based on the bright area image and the dark area image, the method further comprises:

[0020] the average value of each row of pixels in the bright area image and the dark area image is calculated respectively to obtain a bright area row average pixel and a dark area row average pixel.

[0021] Further, the constructing the denoising model based on the bright area image and the dark area image comprises:

[0022] The denoising model is constructed by fitting the bright area average pixel, the dark area average pixel, the bright area row average pixel and the dark area row average pixel.

[0023] According to another aspect of the present application, a denoising device for image data is provided, comprising:

[0024] An acquisition module is configured to acquire standard image data collected under different environments;

[0025] A correction module is configured to perform geometric correction on the standard image data to obtain corrected image data, and perform area division based on the corrected image data to obtain a bright area image and a dark area image;

[0026] A denoising module is configured to construct a denoising model based on the bright area image and the dark area image, and perform denoising on target image data to be denoised based on the denoising model to obtain denoised image data.

[0027] Further,

[0028] The correction module is specifically configured to perform stripe conversion on the standard image data by a line-by-line correction method to obtain standard image data in a regular strip structure, calculate mean square errors between a first row of pixels of the standard image data in the regular strip structure and each row of pixels by a sliding window method under the condition that the first row of pixels is determined as reference pixels, and correct the standard image data in the regular strip structure according to offset pixels corresponding to the minimum mean square error to obtain the corrected image data.

[0029] Further, the device further comprises:

[0030] An elimination module is configured to eliminate random noise from the standard image data by using minimum pixel values of each row of the standard image data as standard image data to be geometrically corrected.

[0031] Further,

[0032] The correction module is specifically configured to perform area division on the corrected image data by using at least one of a histogram peak value, local gray fluctuation, pixel density distribution characteristics and a global adaptive threshold algorithm to obtain a bright area image and a dark area image.

[0033] Further, the device further comprises:

[0034] The computing module is configured to perform average value calculation on the region pixels of the bright region image and the dark region image respectively to obtain bright region average pixels and dark region average pixels.

[0035] Further, the computing module is further configured to perform average value calculation on the row pixels in the bright region image and the dark region image respectively to obtain bright region row average pixels and dark region row average pixels.

[0036] Further, the denoising module is specifically configured to perform fitting through the bright region average pixels, the dark region average pixels, the bright region row average pixels and the dark region row average pixels to construct the denoising model.

[0037] According to another aspect of the present application, a storage medium is provided, and the storage medium stores at least one executable instruction, and the executable instruction causes a processor to perform operations corresponding to the image data denoising method.

[0038] According to still another aspect of the present application, a terminal is provided, and the terminal comprises a processor, a memory, a communication interface and a communication bus, and the processor, the memory and the communication interface complete communication with each other through the communication bus.

[0039] The memory is configured to store at least one executable instruction, and the executable instruction causes the processor to perform operations corresponding to the image data denoising method.

[0040] Through the above technical solutions, the technical solutions provided by the embodiments of the present application have at least the following advantages:

[0041] The present application provides an image data denoising method and device. Compared with the prior art, the embodiments of the present application acquire standard image data collected under different environments; perform geometric correction on the standard image data to obtain corrected image data, and perform region division based on the corrected image data to obtain a bright region image and a dark region image; construct a denoising model based on the bright region image and the dark region image, and when target image data to be denoised is acquired, perform denoising on the target image data based on the denoising model to obtain denoised image data, so as to output a denoised image with unified brightness and darkness, corrected strips and enhanced edges, significantly improve the visual quality and the robustness of subsequent image recognition processing, and thus achieve the purpose of accurate and effective denoising.

[0042] The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application, the specific embodiments of the present application can be implemented in accordance with the content of the description, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described. BRIEF DESCRIPTION OF DRAWINGS

[0043] Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The accompanying drawings are included to provide a better understanding of the preferred embodiments, and are not intended to constrain the application. Moreover, in the drawings, the same reference numerals designate the same components throughout the several views. In the drawings:

[0044] Figure 1 A flow chart of a method for removing noise from image data is shown according to an embodiment of the present application;

[0045] Figure 2 A comparison chart of effects of calibrating image geometry is shown according to an embodiment of the present application;

[0046] Figure 3 A comparison chart of effects of removing noise is shown according to an embodiment of the present application;

[0047] Figure 4 A flow chart of a system for removing noise from a noise image by fitting pixel values is shown according to an embodiment of the present application;

[0048] Figure 5 A block diagram of a device for removing noise from image data is shown according to an embodiment of the present application;

[0049] Figure 6 A structure diagram of a terminal is shown according to an embodiment of the present application. DETAILED DESCRIPTION

[0050] Exemplary embodiments of the present disclosure will be described more fully hereinafter with reference to the accompanying drawings. While example embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be embodied in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the present disclosure to those skilled in the art.

[0051] A method for removing noise from image data is provided according to an embodiment of the present application, as shown in the figure, the method comprises: Figure 1

[0052] 101, obtaining standard image data collected under different environments.

[0053] ​In the embodiment of the present application, the current execution end as the denoising execution subject of image data can be a terminal processor, and can also be a server to obtain standard image data collected in different environments. The standard image data as a calibration image is obtained by shooting under uniform illumination for image acquisition devices (such as cameras or infrared sensing devices), and one corresponding standard image data can be obtained for different environments and different sensors. At this time, as a calibration image, the collected standard image data contains relatively stable gray scale distribution characteristics and can cover the entire field of view of the image acquisition device. In addition, taking the standard image data as a calibration reference can reflect the brightness inconsistency of the imaging system on each column or each row of pixels due to device gain offset, electrical response difference or temperature non-uniformity.

[0054] It should be noted that the shooting objects of the standard image data in the embodiment of the present application include but are not limited to landscapes, objects, people, etc., to adapt to the denoising purposes of different images, and the embodiment of the present application is not limited.

[0055] 102. Geometrically correcting the standard image data to obtain corrected image data, and performing region division based on the corrected image data to obtain a bright region image and a dark region image.

[0056] In the embodiment of the present application, after the current execution end obtains the standard image data, the standard image data is processed by geometric correction to unify the rows and columns, that is, the corrected image data is obtained after geometric correction, so that the strip noise in the standard image data can be arranged in the correct direction, as shown in the following figure, so as to establish a more accurate denoising model subsequently. For geometric correction, the geometric correction can be performed in a row-by-row manner according to the arrangement mode of the specific image sensor to convert the non-structural stripes in the standard image data into a regular strip structure to achieve the purpose of correction. Figure 2

[0057] In addition, after obtaining the corrected image data, the current execution end performs region division on the corrected image data according to the bright-dark distribution to obtain a bright region image and a dark region image. The standard image data can be divided into a bright region image and a dark region image by means of statistical histogram peak value, local gray scale fluctuation or pixel density distribution characteristics.

[0058] 103. Constructing a denoising model based on the bright region image and the dark region image, and when obtaining target image data to be denoised, performing denoising on the target image data based on the denoising model to obtain denoised image data.

[0059] ​In the embodiments of the present application, the current execution end constructs a denoising model through the bright area image and the dark area image. At this time, the input variable of the denoising model is represented as the pixel response value before correction, and the output variable is represented as the pixel response value after correction, as shown in FIG. 13. The denoising model can be represented as: Figure 3

[0060]

[0061]

[0062] wherein, the pixel response value after correction is represented as the output variable is represented as the pixel response value before correction, the bright area average pixel is represented as the dark area average pixel is represented as the bright area average pixel is represented as the dark area average pixel is represented as

[0063] In another embodiment of the present application, in order to further limit and illustrate, the step of performing geometric correction on the standard image data to obtain corrected image data includes:

[0064] performing stripe conversion on the standard image data by a row-by-row correction method to obtain standard image data in a regular strip structure;

[0065] determining a first row of pixels of the standard image data in the regular strip structure as reference pixels, and calculating the mean square error between the first row of pixels and each row of pixels by a sliding window method under the condition that the first row of pixels is determined as the reference pixels;

[0066] correcting the standard image data in the regular strip structure according to the offset pixel corresponding to the minimum mean square error to obtain the corrected image data.

[0067] ​​​​In order to achieve the purpose of eliminating image noise by geometric correction, the current execution end first converts the standard image data into regular strip structure by row-by-row correction when performing geometric correction. Specifically, the current execution end can convert the non-structural stripes in the original image into regular strip structure according to the specific image sensor arrangement by using row-by-row correction. At this time, the row-by-row correction can be achieved by extracting the target contour (such as the document edge) in the image, calculating the minimum circumscribed rectangle and applying four-point perspective transformation to realize row alignment. It can also adjust the pixel position row by row through a polynomial model (such as Brown-Conrady model) for edge stretching or compression caused by lens distortion, and the present application embodiment is not limited specifically. Further, the current execution end determines the first row of pixels of the regular strip structure of the standard image data as the reference pixel, and calculates the mean square error of the first row of pixels and each row of pixels by sliding window method to obtain the best pixel offset value by minimizing the mean square error. At this time, the regular strip structure of the standard image data is corrected according to the offset pixel corresponding to the minimum mean square error, and the corrected image data is obtained, which can be represented as:

[0068] ;

[0069] ;

[0070] wherein, represents the jthpixel of the row number i, is the j-i pixel of the first row, represents the minimum mean square error (MSE) offset pixel value of the row number i. In another embodiment of the present application, in order to further limit and illustrate, before the step of performing geometric correction on the standard image data to obtain corrected image data, the method further comprises:

[0071] Random noise elimination is performed on the standard image data by the minimum pixel value of each row in the standard image data as the standard image data to be geometrically corrected.

[0072] In order to eliminate the influence of random noise and improve the denoising efficiency of image data, the current execution end eliminates the random noise of the standard image data before performing geometric correction. Specifically, the random noise of the standard image data can be eliminated by the minimum pixel value of each row in the standard image data, which avoids the interference of subsequent mean square error calculation and makes the geometric correction result unstable, which is represented as:

[0073]

[0074] ;​​

[0075] wherein, represents the row number, i represents the column number, represents the minimum pixel value of the row.

[0076] In another embodiment of the present application, in order to further limit and illustrate, the step of dividing the region based on the corrected image data to obtain a bright region image and a dark region image comprises:

[0077] The region of the corrected image data is divided by at least one of the histogram peak, the local gray scale fluctuation, the pixel density distribution feature, and the global adaptive threshold algorithm to obtain a bright region image and a dark region image.

[0078] In order to realize the construction of the denoising model based on the bright and dark region division method, thereby improving the image removal efficiency, when the region is divided, the current execution end can adopt at least one of the histogram peak, the local gray scale fluctuation, the pixel density distribution feature, and the global adaptive threshold algorithm to divide the region of the corrected image data. Among them, the region division of the histogram peak is to divide the region based on the gray value interval with the most number of pixels in the histogram, which reflects the most concentrated brightness or color distribution feature in the image; the region division of the local gray scale fluctuation is to combine pixel density distribution, edge detection and region growth technology to divide the region; the region division of the pixel density distribution feature is to divide the image into different regions by analyzing the statistical characteristics of pixel gray scale or color; the global adaptive threshold algorithm is to dynamically adjust the threshold value of the local region, and the region is segmented according to the light or background complex scene. The local threshold value can be calculated based on the pixel neighborhood statistics (such as mean, Gaussian weighting) to realize segmentation. In addition, when the global adaptive threshold algorithm is used in the bright and dark domain division process, the Otsu method can be selected, or the K-means algorithm based on the distribution of clustering centers is used to realize, so as to generate a bright and dark mask image. The mask is used to locate the typical pixel set in different regions, and the embodiments of the present application are not limited.

[0079] In another embodiment of the present application, in order to further limit and illustrate, before the step of constructing the denoising model based on the bright region image and the dark region image, the method further comprises:

[0080] The average values of the region pixels of the bright region image and the dark region image are calculated respectively to obtain a bright region average pixel and a dark region average pixel.

[0081] In order to construct a more effective denoising model, the current execution end calculates the average values corresponding to the region pixels of the bright region image and the dark region image respectively in advance, which is represented as:

[0082] ;

[0083] ;

[0084] wherein, is the average pixel of the bright region, is the average pixel of the dark region, is the pixel set of the bright region, is the pixel set of the dark region.

[0085] In another embodiment of the present application, in order to further limit and illustrate, the method further comprises, before the step of constructing the denoising model based on the bright region image and the dark region image:

[0086] respectively calculating the average value of each row of pixels in the bright region image and the dark region image to obtain the bright region row average pixel and the dark region row average pixel.

[0087] In order to construct a more effective denoising model, the current execution end calculates the average value of each row of pixels in the bright region image and the dark region image in advance, which is expressed as:

[0088] ;

[0089] ;

[0090] wherein, is the bright region row average pixel, is the dark region row average pixel, is the pixel set of the bright region, is the pixel set of the dark region.

[0091] In another embodiment of the present application, in order to further limit and illustrate, the step of constructing the denoising model based on the bright region image and the dark region image comprises:

[0092] fitting through the bright region average pixel, the dark region average pixel, the bright region row average pixel and the dark region row average pixel to construct the denoising model.

[0093] In order to achieve the purpose of constructing a denoising model for image denoising, the current execution end, when constructing the denoising model, specifically, constructs a mathematical model expressing the row brightness mapping relationship between the standard image data and the to-be-denoised image data through two-point fitting, such as the algorithm flowchart for removing noise of the image system through pixel value fitting as shown in Figure 4 , and the denoising model is expressed as:

[0094] ;

[0095] ;

[0096] wherein, is a corrected pixel response value, is an output variable representing a pixel response value before correction, is a bright field row average pixel, is a dark field row average pixel, is a bright area average pixel, is a dark area average pixel, and k and b are fitting coefficients.

[0097] It should be noted that after the current execution end constructs the denoising model, the reference brightness for describing how to eliminate the systematic noise of the image based on the pixel in the standard image data is realized, so as to form the purpose of compensating for the systematic noise.

[0098] The embodiment of the present application provides a denoising method of image data. Compared with the prior art, the embodiment of the present application acquires standard image data collected under different environments; geometrically corrects the standard image data to obtain corrected image data, and performs regional division based on the corrected image data to obtain a bright area image and a dark area image; constructs a denoising model based on the bright area image and the dark area image, and when target image data to be denoised is acquired, denoises the target image data based on the denoising model to obtain denoised image data, so as to output a denoised image after brightness and darkness unification, strip correction and edge enhancement, significantly improve the visual quality and the robustness of subsequent image recognition processing, and thus realize the purpose of accurate and effective denoising.

[0099] Further, as an implementation of the method shown in the above Figure 1 The embodiment of the present application provides a denoising device of image data, as shown in the above Figure 5 The device comprises:

[0100] An acquisition module 21 is configured to acquire standard image data collected under different environments.

[0101] A correction module 22 is configured to geometrically correct the standard image data to obtain corrected image data, and perform regional division based on the corrected image data to obtain a bright area image and a dark area image.

[0102] A denoising module 23 is configured to construct a denoising model based on the bright area image and the dark area image, and when target image data to be denoised is acquired, denoise the target image data based on the denoising model to obtain denoised image data.

[0103] Further,

[0104] The correction module is specifically configured to perform stripe conversion on the standard image data by a row-by-row correction manner to obtain standard image data in a regular strip structure; under the condition that a first row of pixels of the standard image data in the regular strip structure is determined as reference pixels, the correction module is further configured to calculate mean square errors between the first row of pixels and each row of pixels by a sliding window manner; and the correction module is further configured to correct the standard image data in the regular strip structure according to offset pixels corresponding to the minimum mean square error to obtain the corrected image data.

[0105] Further, the device further comprises:

[0106] The elimination module is configured to eliminate random noise in the standard image data by using minimum pixel values of each row of the standard image data as the standard image data to be subjected to geometric correction.

[0107] Further,

[0108] The correction module is specifically configured to perform region division on the corrected image data by using at least one of a histogram peak value, local grayscale fluctuation, pixel density distribution characteristics and a global adaptive threshold algorithm to obtain a bright region image and a dark region image.

[0109] Further, the device further comprises:

[0110] The calculation module is configured to perform average value calculation on region pixels of the bright region image and the dark region image respectively to obtain a bright region average pixel and a dark region average pixel.

[0111] Further, the calculation module is further configured to perform average value calculation on each row of pixels in the bright region image and the dark region image respectively to obtain a bright region row average pixel and a dark region row average pixel.

[0112] Further, the denoising module is specifically configured to perform fitting by using the bright region average pixel, the dark region average pixel, the bright region row average pixel and the dark region row average pixel to construct the denoising model.

[0113] The embodiment of the present application provides a kind of image data denoising device, compared with prior art, the embodiment of the present application is acquired by standard image data collected under different environments;The standard image data is geometrically corrected, and correction image data is obtained, and region division is carried out based on the correction image data, and bright area image and dark area image are obtained;Based on the bright area image and the dark area image, a denoising model is constructed, and when the target image data to be denoised is acquired, the target image data is denoised based on the denoising model, and the image data after denoising is obtained, to output the denoising image after bright-dark uniformity, strip correction, edge enhancement, significantly improve the robustness of visual quality and subsequent image recognition processing, to realize the purpose of accurate and effective denoising.

[0114] According to an embodiment of the present application, a storage medium is provided, and the storage medium stores at least one executable instruction.

[0115] Figure 6 A structure schematic diagram of a terminal according to an embodiment of the present application is shown, and the specific implementation of the terminal is not limited in the specific embodiments of the present application.

[0116] As shown in Figure 6 The terminal can include: processor 302, communications interface 304, memory 306, and communication bus 308.

[0117] Among them: processor 302, communications interface 304, and memory 306 complete mutual communication through communication bus 308.

[0118] The communications interface 304 is used to communicate with network elements of other devices, such as clients or other servers.

[0119] The processor 302 is used to execute the program 310, and specifically can execute the related steps in the above image data denoising method embodiments.

[0120] Specifically, the program 310 can include program code, and the program code includes computer operation instructions.

[0121] The processor 302 can be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to perform the embodiments of the present application. The terminal includes one or more processors, which can be the same type of processor, such as one or more CPUs; or can be different types of processors, such as one or more CPUs and one or more ASICs.

[0122] The memory 306 is used to store programs 310. The memory 306 can include a high-speed RAM memory, and can also include a non-volatile memory, such as at least one disk memory.

[0123] The programs 310 can be specifically used to make the processor 302 perform the following operations:

[0124] Obtaining standard image data collected in different environments;

[0125] Geometrically correcting the standard image data to obtain corrected image data, and performing region division based on the corrected image data to obtain a bright region image and a dark region image;

[0126] Constructing a denoising model based on the bright region image and the dark region image, and when obtaining target image data to be denoised, performing denoising on the target image data based on the denoising model to obtain denoised image data.

[0127] Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present application can be realized by a general computing device, which can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and can be realized by program codes executable by the computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in different order, or they can be manufactured into individual integrated circuit modules, or multiple modules or steps can be manufactured into a single integrated circuit module. Thus, the present application is not limited to any specific hardware and software combination.

[0128] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method of denoising image data, characterized by, The method comprises the following steps: acquiring standard image data collected under different environments; performing geometric correction on the standard image data to obtain corrected image data, and performing region division based on the corrected image data to obtain a bright region image and a dark region image; constructing a denoising model based on the bright region image and the dark region image, and performing denoising on target image data based on the denoising model to obtain denoised image data when the target image data is acquired; the geometric correction on the standard image data to obtain corrected image data comprises: performing stripe conversion on the standard image data by a row-by-row correction method to obtain standard image data in a regular strip structure; under the condition that the first row of pixels of the standard image data in the regular strip structure is determined as a reference pixel, calculating the mean square error of the first row of pixels and each row of pixels by a sliding window method; correcting the standard image data in the regular strip structure according to the offset pixel corresponding to the minimum mean square error to obtain the corrected image data; before the denoising model is constructed based on the bright region image and the dark region image, the method further comprises: respectively calculating the average values of each row of pixels in the bright region image and the dark region image to obtain bright region row average pixels and dark region row average pixels; the construction of the denoising model based on the bright region image and the dark region image comprises: fitting the bright region average pixels, the dark region average pixels, the bright region row average pixels and the dark region row average pixels to construct the denoising model, wherein the denoising model is characterized by a two-point fitting method to construct a mathematical model representing the brightness mapping relationship between the standard image data and the target image data, and is expressed as: ； ； wherein, is the corrected pixel response value, is the output variable, representing the pixel response value before correction, is the bright region average pixel, is the dark region average pixel, is the bright region average pixel, is the dark region average pixel, k, b are fitting coefficients.

2. The method of claim 1, wherein, before the geometric correction on the standard image data to obtain corrected image data, the method further comprises: performing random noise elimination on the standard image data by using the minimum pixel value of each row of pixels in the standard image data as the standard image data to be geometrically corrected.

3. The method of claim 1, wherein, the region division based on the corrected image data to obtain the bright region image and the dark region image comprises: performing region division on the corrected image data by at least one of a histogram peak value, a local gray level fluctuation, a pixel density distribution feature and a global adaptive threshold algorithm to obtain the bright region image and the dark region image.

4. The method of claim 3, wherein, before the denoising model is constructed based on the bright region image and the dark region image, the method further comprises: respectively calculating the average values of the region pixels of the bright region image and the dark region image to obtain bright region average pixels and dark region average pixels.

5. An apparatus for de-noising image data, characterized by, The method comprises the following steps: an acquisition module, configured to acquire standard image data collected under different environments; a correction module, configured to perform geometric correction on the standard image data to obtain corrected image data, and perform region division based on the corrected image data to obtain a bright region image and a dark region image; The denoising module is configured to construct a denoising model based on the bright area image and the dark area image, and to denoise target image data based on the denoising model to obtain denoised image data when the target image data is acquired. The correction module is specifically configured to perform stripe conversion on the standard image data by a row-by-row correction manner to obtain standard image data in a regular strip structure; under the condition that a first row of pixels of the standard image data in the regular strip structure is determined as reference pixels, to calculate mean square errors between the first row of pixels and each row of pixels by a sliding window manner; and to correct the standard image data in the regular strip structure according to offset pixels corresponding to the minimum mean square error to obtain the corrected image data. The device further comprises: The calculation module is configured to calculate average values of each row of pixels in the bright area image and the dark area image respectively to obtain bright field row average pixels and dark field row average pixels. The denoising module is specifically configured to perform fitting on the bright area average pixels, the dark area average pixels, the bright field row average pixels and the dark field row average pixels to construct the denoising model, wherein the denoising model is expressed as a mathematical model for representing a mapping relationship between the standard image data and the target image data row brightness by two-point fitting, and is expressed as: ； ； wherein, is the corrected pixel response value, is the output variable, representing the uncorrected pixel response value, is the bright region average pixel, is the dark region average pixel, is the bright region average pixel, is the dark region average pixel, k, b are fitting coefficients.

6. A computer readable storage medium having stored thereon computer programs / instructions, characterized in that, The computer program / instruction is executed by the processor to implement the steps of the method of claim 1.

7. A computer device comprising a memory, a processor, and a computer program stored on the memory, wherein the computer program comprises instructions that, when executed by the processor, cause the processor to perform the method of any one of claims 1-6. The processor executes the computer program to implement the steps of the method of claim 1.

Images