CN116894797A - Compression denoising method and system for high-dynamic bayer image - Google Patents
Compression denoising method and system for high-dynamic bayer image Download PDFInfo
- Publication number
- CN116894797A CN116894797A CN202311162972.6A CN202311162972A CN116894797A CN 116894797 A CN116894797 A CN 116894797A CN 202311162972 A CN202311162972 A CN 202311162972A CN 116894797 A CN116894797 A CN 116894797A
- Authority
- CN
- China
- Prior art keywords
- image
- high dynamic
- dynamic range
- denoising
- pixel value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000007906 compression Methods 0.000 title claims abstract description 44
- 230000006835 compression Effects 0.000 title claims abstract description 44
- 230000009466 transformation Effects 0.000 claims description 47
- 238000006243 chemical reaction Methods 0.000 claims description 28
- 238000012545 processing Methods 0.000 claims description 16
- 230000006837 decompression Effects 0.000 claims description 9
- 239000000758 substrate Substances 0.000 claims 1
- 239000004973 liquid crystal related substance Substances 0.000 abstract 2
- 238000012937 correction Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 230000001419 dependent effect Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 238000007796 conventional method Methods 0.000 description 2
- 238000013144 data compression Methods 0.000 description 2
- 230000000750 progressive effect Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000001131 transforming effect Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 229910044991 metal oxide Inorganic materials 0.000 description 1
- 150000004706 metal oxides Chemical class 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
Landscapes
- Image Processing (AREA)
Abstract
The application provides a compression denoising method and a system for a high-dynamic bayer image, wherein the method comprises the following steps: acquiring a high dynamic range image of a bayer domain; by passing throughThe method comprises the steps of compressing pixels in a high dynamic range image, and simultaneously converting noise related to signals in the high dynamic range image into noise unrelated to the signals to obtain a low-bit wide compressed image; wherein, the liquid crystal display device comprises a liquid crystal display device,a pixel value expressed as the i-th pixel in the high dynamic range image, a, b, and c are constants; and denoising the low-bit-width compressed image. The application can ensure the denoising performance of the denoising algorithm.
Description
Technical Field
The application relates to the technical field of Image Signal Processing (ISP), in particular to a compression denoising method and a system for a high-dynamic Bayer image.
Background
In the ISP processing of the HDR (high dynamic range), in order to preserve a larger dynamic range, the bit width of the image data processed by the hdr_combine (high dynamic range combining) module is very wide, and often reaches a data bit width of 14 bits to 20 bits, or even 24 bits. If the data with wide bit width, i.e. the HDR data, is directly used for denoising, the line buffer (linebuffer) required by the denoising module is very large, and the line buffer is usually required to reach about 10 to 20 lines, so that the chip is required to use a larger data bit width, thereby leading to a rapid increase of the area of the chip.
On the other hand, if the general data compression algorithm is directly adopted to perform bit width compression of the wide dynamic data, such as log (log) transformation compression and Sqrt compression of the image, the linearity of the wide dynamic data is destroyed, and thus the performance of the denoising algorithm is reduced due to the fact that the noise model of the bayer data cannot be accurately estimated.
Disclosure of Invention
In order to solve the problems, the application provides a compression denoising method and a system for a high-dynamic bayer image, which are implemented byThe pixels in the high dynamic range image are processed, so that the denoising performance of a denoising algorithm can be ensured while the increase of the chip area is avoided.
In a first aspect, the present application provides a compression denoising method for a high dynamic bayer image, including:
acquiring a high dynamic range image of a bayer domain;
by passing throughThe method comprises the steps of compressing pixels in a high dynamic range image, and simultaneously converting noise related to signals in the high dynamic range image into noise unrelated to the signals to obtain a low-bit wide compressed image; wherein (1)>A pixel value expressed as the i-th pixel in the high dynamic range image, a, b, and c are constants;
and denoising the low-bit-width compressed image.
Alternatively, the process may be carried out in a single-stage,the GAT conversion formula is combined with the maximum pixel value and the minimum pixel value in the denoising image, and the denoising image is obtained by processing the high dynamic range image through the GAT conversion formula.
Alternatively, the GAT transformation formula isWherein, the method comprises the steps of, wherein,and is a scale factor,/->Is the standard deviation of gaussian noise;
=/>，/>minimum pixel value in denoised image obtained by GAT transformation formula for high dynamic range image,/and>is the difference between the maximum pixel value and the minimum pixel value in the denoised image.
Optionally byThe step of compressing pixels in the high dynamic range image and simultaneously converting noise related to signals in the high dynamic range image into noise unrelated to the signals to obtain a low-order wide compressed image comprises the following steps:
high dynamic signal in high dynamic range imageThrough->To obtain an intermediate signal Y, performing an evolution operation on Y to obtain an intermediate signal Z, and performing a subtraction operation on Z and c to obtain a final output N-bits transformed signal->And obtaining the low-bit-width compressed image.
Optionally, the method further comprises: and decompressing the low-bit-width compressed image after denoising processing to obtain a denoising image with a high dynamic range.
Optionally, the step of decompressing the denoised low-order wide compressed image to obtain a denoised image with a high dynamic range includes:
according toOr an Anscombe inverse transformation formula, determining inverse transformation parameters, and decompressing the denoised low-bit-width compressed image through the inverse transformation parameters to obtain a high-dynamic-range denoised image.
In a second aspect, the present application provides a compression denoising system for a high dynamic bayer image, comprising:
an acquisition module configured to acquire a high dynamic range image of a bayer domain;
a compression conversion module configured to convert byThe method comprises the steps of compressing pixels in a high dynamic range image, and simultaneously converting noise related to signals in the high dynamic range image into noise unrelated to the signals to obtain a low-bit wide compressed image; wherein (1)>A pixel value expressed as the i-th pixel in the high dynamic range image, a, b, and c are constants;
and the denoising module is configured to denoise the low-bit-width compressed image.
Alternatively, the process may be carried out in a single-stage,the method comprises the steps of combining a maximum pixel value and a minimum pixel value in a denoising image by using a GAT conversion formula, and processing the denoising image by using a high dynamic range image by using the GAT conversion formula.
Alternatively, the GAT transformation formula isWherein, the method comprises the steps of, wherein,and is a scale factor,/->Is the standard deviation of gaussian noise;
=/>，/>minimum pixel value in denoised image obtained by GAT transformation formula for high dynamic range image,/and>is the difference between the maximum pixel value and the minimum pixel value in the denoised image.
Optionally, the compression conversion module is further configured to convert the high dynamic signal in the high dynamic range imageThrough the process ofTo obtain an intermediate signal Y, performing an evolution operation on Y to obtain an intermediate signal Z, and performing a subtraction operation on Z and c to obtain a final output N-bits transformed signal->And obtaining the low-bit-width compressed image.
Optionally, the system further comprises:
and the decompression module is configured to decompress the low-bit-width compressed image after the denoising processing to obtain a denoising image with a high dynamic range.
Optionally, the decompression module is further configured toOr an Anscombe inverse transformation formula, determining inverse transformation parameters, and decompressing the denoised low-bit-width compressed image through the inverse transformation parameters to obtain a high-dynamic-range denoised image.
In a third aspect, the present application provides an electronic device comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of the above.
In a fourth aspect, the present application provides a chip comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of the above.
In a fifth aspect, the application provides a computer readable storage medium storing computer instructions which when executed by a processor perform a method of any of the above.
The embodiment of the application provides a compression denoising method and a system for a high-dynamic Bayer image, which are implemented by the following steps ofThe pixels in the high dynamic range image are processed, so that the high dynamic range image can be compressed, meanwhile, noise related to signals in the high dynamic range image is converted into noise unrelated to the signals, the follow-up denoising module can accurately and efficiently denoise, and the reduction of denoising performance and the loss of image details caused by compression are avoided.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments or the conventional techniques of the present application, the drawings required for the descriptions of the embodiments or the conventional techniques will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to the drawings without inventive effort for those skilled in the art.
FIG. 1 is a schematic flow chart of a compression denoising method of a high-dynamic bayer image according to an embodiment of the present application;
FIG. 2 is a schematic flow chart of an open-mode compression transformation of a high dynamic range image in accordance with an embodiment of the present application;
FIG. 3 is a schematic flow chart of decompressing a denoised low-bit-width compressed image according to an embodiment of the present application;
fig. 4 is a schematic block diagram of a compression denoising system of a high-dynamic bayer image according to an embodiment of the present application.
Detailed Description
In order that the application may be readily understood, a more complete description of the application will be rendered by reference to the appended drawings. Embodiments of the application are illustrated in the accompanying drawings. This application may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the description of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.
First, the proper nouns according to the present application will be explained as follows:
HDR: high Dynamic Range, high dynamic range;
raw: an original image file containing data processed from an image sensor of a digital camera, scanner, or motion picture film scanner;
bayer: a data format of raw;
BLC: black Level Correction, black level correction;
NR: noise Reduction, noise removal;
DRC: dynamic Range Compression, dynamic range compression;
GAT: generalized Anscombe Transform, generalized Anscombe transform;
IAT: inverse Anscombe Transform Anscombe inverse transformation.
In this regard, it is noted that for High Dynamic Range (HDR) technology, the range of brightness that can be discerned by the real scene and human eye can be as high asBut the dynamic range that can be captured by a common image sensor does not exceed + ->This is far less than the dynamic range of a real scene, and has a certain gap from the dynamic range that can be observed by the human eye. Therefore, a common CMOS/CCD image sensor cannot completely present a real scene with large brightness level difference. Meanwhile, in a brighter area in a real scene, a white bright area appears due to overexposure, and a dark area in the scene appears due to underexposure. In order to reproduce a high dynamic range scene, it is generally necessary to expose the same scene multiple times to obtain a plurality of images with different exposure degrees, and then to obtain a high dynamic range image through image synthesis.
In a first aspect, an embodiment of the present application provides a compression denoising method of a high-dynamic bayer image, which is applicable to imaging devices such as digital cameras and any devices equipped with CCD (charge coupled device) or CMOS (complementary metal oxide semiconductor) sensors, which capture images by continuous photon-to-electron, electron-to-voltage, and voltage-to-digital conversion. This process of capturing images is subject to various signal-dependent errors, and the standard approach to modeling these errors is to consider them as poisson-gaussian noise. We markFor clean and noiseless image->For an observed noisy image, then the noise model for the ith noisy pixel is represented as follows:
in the case of the formula (1),is the signal-dependent poisson noise caused by photon sensing,other signal independent gaussian noise. Here, α>0 and is a scale factor which depends on the sensor and the analog gain, +.>Is the standard deviation of gaussian noise.
Based on equation (1), for each clean and noisy image, it is assumed that each pixel is normalized and clipped to have a range of [0,1]An inner value. Thus, the poisson-gaussian noise model can be defined by parametersTo characterize, the noise variance of Y can be expressed as
Wherein the noise parameter of equation (2)This can be obtained by calibrating the gray card at different iso (sensitivities).
Further, a common method of denoising a noise image contaminated with poisson-gaussian noise is to apply the GAT transform, which is the generalized aThe nscombe transform is capable of converting signal-dependent noise in an input image into signal-independent noise. The input image can then be denoised by using existing denoising methods. Specifically, based on equation (2), the GAT transform will every pixelConversion to equation (3) is as follows:
wherein,,the transformation may stabilize the noise of each pixel in the transformed input image to approximate a gaussian distribution with unity variance, i.e. approximately obey +.>Is a gaussian distribution of (c). With this feature, the poisson-Gaussian noise can be more simply denoised, i.e., a general Gaussian denoising device can be applied toAnd obtaining a denoising image D on the transformed image. Then, an Inverse Anscombe (IAT) transform is applied to the denoised image D to obtain an original clean image +.>Is used to estimate the final estimate of (a).
Wherein the direct inverse of the Anscombe transform is given by equation (4),
the exact unbiased inverse of the ascombe transformation is given by equation (5),
the reduced progressive inverse of the ascombe transform is given by equation (6),
wherein,,denoted as->Is the i-th pixel of (c). Thus, the final denoised image becomes +.>. It should be noted that noise parameters are required for both GAT and IAT conversion>。
Referring to fig. 1, in order to solve the problems in the background art, the compression denoising method of the high-dynamic bayer image provided in the present embodiment includes steps S101 to S103:
step S101: a high dynamic range image of the bayer domain is acquired.
Wherein the high dynamic range image of the bayer domain is also referred to as an HDR bayer image. In the present embodiment, the high dynamic range image of the bayer domain is a noisy high dynamic range image.
Step S102: by passing throughAnd compressing pixels in the high dynamic range image, and simultaneously converting noise related to signals in the high dynamic range image into noise irrelevant to the signals to obtain a low-bit-width compressed image.
Wherein,,the pixel value, denoted as the i-th pixel in the high dynamic range image, a, b and c are all constants.
In an alternative embodiment of the present application,the method comprises the steps of combining a maximum pixel value and a minimum pixel value in a denoising image by using a GAT conversion formula, and processing the denoising image by using a high dynamic range image by using the GAT conversion formula.
In an alternative embodiment, the GAT transformation formula isWherein->And is a scale factor,/->Is the standard deviation of gaussian noise;
=/>，/>minimum pixel value in denoised image obtained by GAT transformation formula for high dynamic range image,/and>is the difference between the maximum pixel value and the minimum pixel value in the denoised image.
In the present embodiment, the compression transformation formula based on GAT is after evolutionAnd c.
Specifically, compression based on the Anscombe transform. Taking into account the normalization of the data in the transform domain according to equation (3), let，Thus, anscombe transformation canThe conversion into the following form:
order theThere is a compression transformation formula based on GAT, namely formula (7) as follows:
this turns the GAT transformation problem into computationAnd then subtracting a constant c. The method utilizes the variation characteristic of the open data bit width to compress the data bit width of the high dynamic range image, which is similar to SQRT compression, thereby having the function of Anscombe transformation.
In an alternative embodiment, byThe step of compressing pixels in the high dynamic range image and simultaneously converting noise related to signals in the high dynamic range image into noise unrelated to the signals to obtain a low-order wide compressed image comprises the following steps:
high dynamic signal in high dynamic range imageThrough->To obtain an intermediate signal Y, performing an evolution operation on Y to obtain an intermediate signal Z, and performing a subtraction operation on Z and c to obtain a final output N-bits transformed signal->And obtaining the low-bit-width compressed image.
Specifically, the transformation parameters determined according to equation (7)Acting on the whole transformation flow. With reference to FIG. 2, for a high dynamic signal X of one M-bits in a high dynamic range image, go through +.>The intermediate signal Y is obtained by performing a square root computation of SQRT and obtaining the intermediate signal Z from the LUT table (display lookup table), and the intermediate signal Z is subjected to a subtraction operation with c to obtain the final output N-bits transformed signal->。
By constructing a compression mechanism GAT-COMRESS of wide dynamic data, converting GAT conversion into conversion in a normalized more general SQRT form, namely converting a GAT conversion formula into a compression conversion formula based on GAT, compressing data bit width by means of the characteristic of SQRT data bit width change, and thus, the nature of GAT is not changed, and signal related noise in a high dynamic range image can be converted into signal independent noise, so that the accuracy of a subsequent noise model is ensured.
Step S103: and denoising the low-bit-width compressed image.
In an alternative embodiment, the method further comprises: and decompressing the low-bit-width compressed image after denoising processing to obtain a denoising image with a high dynamic range.
In an alternative embodiment, the step of decompressing the denoised low-bandwidth compressed image to obtain the high dynamic range denoised image includes:
according toOr an Anscombe inverse transformation formula, determining inverse transformation parameters by inverse transformationAnd transforming parameters, and decompressing the low-bit-width compressed image after denoising processing to obtain a denoising image with a high dynamic range.
Specifically, based on the formula (7), for the Anscombe direct inverse transform, letThen there is as formula (8):
for Anscombe progressive inverse transform, letThen there is as formula (9):
it should be noted that, the formula (8) or the formula (9) may also be obtained by transforming the formula (4) or the formula (6), which will not be described in detail.
Transformation parameters determined according to equation (8) or equation (9)Acting on the whole inverse transformation flow. Referring to FIG. 3, specifically, an N-bits compressed denoising signal +.>Through->The addition operation results in an intermediate signal Z which is obtained by calculating the SQUARE function, Y being passed through a +.>（/>) To obtain a final outputM-bits high dynamic signal +.>I.e. a high dynamic range de-noised image. Therefore, based on IAT transformation, a decompression algorithm IAT-DECOMRESS of the high dynamic range image data is constructed and used for accurately restoring the denoised high dynamic data, and the details of the image can be better kept.
The compression denoising method for the high-dynamic Bayer image provided by the application realizes the data bit wide compression of the high-dynamic range image, and simultaneously converts the noise related to the signal in the high-dynamic range image into the noise unrelated to the signal, thereby not only reducing the problem of chip area increase caused by direct denoising of the high-dynamic range image, but also ensuring the accurate conversion of a noise model and realizing the efficient denoising of the high-dynamic range image.
In a second aspect, the present application provides a compression denoising system 200 of a high-dynamic bayer image, referring to fig. 4, the compression denoising system 200 of a high-dynamic bayer image includes:
an acquisition module 201 configured to acquire a high dynamic range image of a bayer domain;
a compression conversion module 202 configured to convert the data byThe method comprises the steps of compressing pixels in a high dynamic range image, and simultaneously converting noise related to signals in the high dynamic range image into noise unrelated to the signals to obtain a low-bit wide compressed image; wherein (1)>A pixel value expressed as the i-th pixel in the high dynamic range image, a, b, and c are constants;
the denoising module 203 is configured to denoise the low-bit-width compressed image. The denoising module is also called a Bayer NR module.
In an alternative embodiment of the present application,combining the maximum in the denoised image by the GAT transformation formulaThe pixel value and the minimum pixel value are obtained, and the denoising image is obtained by processing the high dynamic range image through a GAT conversion formula.
In an alternative embodiment, the GAT transformation formula isWherein->And is a scale factor,/->Is the standard deviation of gaussian noise; />=/>，/>Minimum pixel value in denoised image obtained by GAT transformation formula for high dynamic range image,/and>is the difference between the maximum pixel value and the minimum pixel value in the denoised image.
In an alternative embodiment, the compression conversion module 202 is further configured to convert the high dynamic signal in the high dynamic range imageThrough->To obtain an intermediate signal Y, performing an evolution operation on Y to obtain an intermediate signal Z, and performing a subtraction operation on Z and c to obtain a final output N-bits transformed signal->And obtaining the low-bit-width compressed image.
In an alternative embodiment, the compressed denoising system 200 of the high dynamic bayer image further includes:
and the decompression module is configured to decompress the low-bit-width compressed image after the denoising processing to obtain a denoising image with a high dynamic range.
In an alternative embodiment, the decompression module is further configured toOr an Anscombe inverse transformation formula, determining inverse transformation parameters, and decompressing the denoised low-bit-width compressed image through the inverse transformation parameters to obtain a high-dynamic-range denoised image.
It should be noted that, the image compression total module formed by the conversion module 202 and the compression module 203 includes the GAT-COMRESS described above, and is responsible for compressing the high dynamic image with wide bit width (M-bits) into the image with low bit width (N-bits), and meanwhile, has the function of converting the noise related to the signal into the noise unrelated to the signal, so as to ensure that the subsequent Bayer NR module can accurately and efficiently remove noise, and the degradation of the denoising performance and the loss of image details caused by compression are avoided. Meanwhile, a decompression module is arranged behind the denoising module so as to decompress the denoised low-bit-width (N-bits) image into a wide-bit-width (M-bits) high-dynamic-range image, and the denoised high-dynamic-range image is accurately recovered, so that the high-dynamic-range denoising image is obtained.
Wherein, the decompression module contains IAT-DECOMRESS in the above; m can be 14, 16, 18, 20, 24 equivalent, N can be 10, 12, 14, 16 equivalent according to the number of the exposure frames and the actual data bit width of the exposure frames, and satisfies the following conditions。
Specifically, taking a two-frame long-short exposure image with a data bit width of 12 bits as an example, the compression denoising process of the high dynamic range image is described. Wherein, noise parameters of the long exposure image and the short exposure image are [ ]) All already in laboratory ringThe context label is completed.
The long exposure image and the short exposure image of two frames are input to hdr_composite, and output as a high-dynamic image with a data bit width of 16 bits. According to noise parameters of image) Calculating transformation parameter +.>And IAT-DECOMRESS inverse transform parameter +.>. And (3) GAT-COMRESS is carried out on the high-dynamic image with the data bit width of 16 bits, and the data bit width after compression is 12 bits. The Bayer NR denoising is performed on the 12-bit compressed data, and the high dynamic range image is converted from signal-dependent noise to signal-independent noise after the Bayer NR denoising is performed, so that the noise is directly removed by using a classical denoising method. Next, the de-noised image is subjected to an IAT-DECOMRESS inverse transform, and the 12-bit image data is decompressed to a 16-bit high-motion image. The 16-bit high-motion image continues with the processing of subsequent blocks including, but not limited TO, dynamic Range Compression (DRC), demosaicing (DE-MOSAIC), COLOR CORRECTION (COLOR CORRECTION), GAMMA CORRECTION (GAMMA CORRECTION), and COLOR space conversion (RGB-TO-YCC).
In the above process, if 13 lines of linebuffers are used for Bayer NR denoising, and 16bit to 12bit data compression is performed, the linebuffer area of 4bit 13/16bit 13=25% can be approximately saved. And the GAT-COMRESS ensures the stable variance transformation of the noise model, and the denoising effect is not reduced in performance and detail loss caused by data bit width compression.
In a third aspect, the present application provides an electronic device comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of the first aspect.
In a fourth aspect, the present application provides a chip comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of the first aspect.
In a fifth aspect, the application provides a computer readable storage medium storing computer instructions which, when executed by a processor, implement the method of the first aspect.
In the description of the present specification, reference to the terms "some embodiments," "other embodiments," "desired embodiments," and the like, means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. In this specification, schematic descriptions of the above terms do not necessarily refer to the same embodiment or example.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing examples illustrate only a few embodiments of the application, which are described in detail and are not to be construed as limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of protection of the present application is to be determined by the appended claims.
Claims (15)
1. A method for compression denoising of a high dynamic bayer image, comprising:
acquiring a high dynamic range image of a bayer domain;
by passing throughThe method comprises the steps of compressing pixels in a high dynamic range image, and simultaneously converting noise related to signals in the high dynamic range image into noise unrelated to the signals to obtain a low-bit wide compressed image; wherein (1)>A pixel value expressed as the i-th pixel in the high dynamic range image, a, b, and c are constants;
and denoising the low-bit-width compressed image.
2. The method of claim 1, wherein the step of determining the position of the substrate comprises,the method comprises the steps of combining a maximum pixel value and a minimum pixel value in a denoising image by using a GAT conversion formula, and processing the denoising image by using a high dynamic range image by using the GAT conversion formula.
3. The method of claim 2, wherein the GAT transformation formula isWherein->And is a scale factor,/->Is the standard deviation of gaussian noise;
=/>，/>minimum pixel value in denoised image obtained by GAT transformation formula for high dynamic range image,/and>is the difference between the maximum pixel value and the minimum pixel value in the denoised image.
4. A method according to claim 3, characterized by the fact that, byThe step of compressing pixels in the high dynamic range image and simultaneously converting noise related to signals in the high dynamic range image into noise unrelated to the signals to obtain a low-order wide compressed image comprises the following steps:
high dynamic signal in high dynamic range imageThrough->To obtain an intermediate signal Y, performing an evolution operation on Y to obtain an intermediate signal Z, and performing a subtraction operation on Z and c to obtain a final output N-bits transformed signal->And obtaining the low-bit-width compressed image.
5. The method according to any one of claims 1 to 4, further comprising: and decompressing the low-bit-width compressed image after denoising processing to obtain a denoising image with a high dynamic range.
6. The method of claim 5, wherein the step of decompressing the denoised low-bandwidth compressed image to obtain the high dynamic range denoised image comprises:
according toOr an Anscombe inverse transformation formula, determining inverse transformation parameters, and decompressing the denoised low-bit-width compressed image through the inverse transformation parameters to obtain a high-dynamic-range denoised image.
7. A compression denoising system of a high dynamic bayer image, comprising:
an acquisition module configured to acquire a high dynamic range image of a bayer domain;
a compression conversion module configured to convert byThe method comprises the steps of compressing pixels in a high dynamic range image, and simultaneously converting noise related to signals in the high dynamic range image into noise unrelated to the signals to obtain a low-bit wide compressed image; wherein (1)>A pixel value expressed as the i-th pixel in the high dynamic range image, a, b, and c are constants;
and the denoising module is configured to denoise the low-bit-width compressed image.
8. The system of claim 7, wherein the system further comprises a controller configured to control the controller,the method comprises the steps of combining a maximum pixel value and a minimum pixel value in a denoising image by using a GAT conversion formula, and processing the denoising image by using a high dynamic range image by using the GAT conversion formula.
9. According to claim 8Is characterized by the fact that the GAT transformation formula is as followsWherein->And is a scale factor,/->Is the standard deviation of gaussian noise;
=/>，/>minimum pixel value in denoised image obtained by GAT transformation formula for high dynamic range image,/and>is the difference between the maximum pixel value and the minimum pixel value in the denoised image.
10. The system of claim 9, wherein the compression conversion module is further configured to convert the high dynamic signal in the high dynamic range image to a high dynamic signalThrough->To obtain an intermediate signal Y, performing an evolution operation on Y to obtain an intermediate signal Z, and performing a subtraction operation on Z and c to obtain a final output N-bits transformed signal->And obtaining the low-bit-width compressed image.
11. The system according to any one of claims 7 to 10, wherein the system further comprises:
and the decompression module is configured to decompress the low-bit-width compressed image after the denoising processing to obtain a denoising image with a high dynamic range.
12. The system of claim 11, wherein the decompression module is further configured to, according toOr an Anscombe inverse transformation formula, determining inverse transformation parameters, and decompressing the denoised low-bit-width compressed image through the inverse transformation parameters to obtain a high-dynamic-range denoised image.
13. An electronic device, characterized in that the electronic device comprises:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 6.
14. A chip, the chip comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 6.
15. A computer readable storage medium storing computer instructions which, when executed by a processor, implement the method of any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311162972.6A CN116894797A (en) | 2023-09-11 | 2023-09-11 | Compression denoising method and system for high-dynamic bayer image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311162972.6A CN116894797A (en) | 2023-09-11 | 2023-09-11 | Compression denoising method and system for high-dynamic bayer image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116894797A true CN116894797A (en) | 2023-10-17 |
Family
ID=88315271
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311162972.6A Pending CN116894797A (en) | 2023-09-11 | 2023-09-11 | Compression denoising method and system for high-dynamic bayer image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116894797A (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104751415A (en) * | 2013-12-31 | 2015-07-01 | 展讯通信（上海）有限公司 | Image denoising and enhancing method and device and image processing system |
| CN105144232A (en) * | 2014-03-25 | 2015-12-09 | 展讯通信（上海）有限公司 | Methods and systems for denoising images |
-
2023
- 2023-09-11 CN CN202311162972.6A patent/CN116894797A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104751415A (en) * | 2013-12-31 | 2015-07-01 | 展讯通信（上海）有限公司 | Image denoising and enhancing method and device and image processing system |
| CN105144232A (en) * | 2014-03-25 | 2015-12-09 | 展讯通信（上海）有限公司 | Methods and systems for denoising images |
Non-Patent Citations (1)
| Title |
|---|
| JAESEOK BYUN ET AL.: "Learning Blind Pixelwise Affine Image Denoiser With Single Noisy Images", IEEE SIGNAL PROCESSING LETTERS, vol. 27, pages 1105 - 1109, XP011798938, DOI: 10.1109/LSP.2020.3002652 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Faraji et al. | CCD noise removal in digital images | |
| US11127117B2 (en) | Information processing method, information processing apparatus, and recording medium | |
| JP5144202B2 (en) | Image processing apparatus and program | |
| US9129403B2 (en) | Method and system for generating enhanced images | |
| JP6054585B2 (en) | Dynamic color shading correction | |
| US20100141804A1 (en) | Image data processing method by reducing image noise, and camera integrating means for implementing said method | |
| US8169500B2 (en) | Dynamic range compression apparatus, dynamic range compression method, computer-readable recording medium, integrated circuit, and imaging apparatus | |
| JP2009512290A (en) | Image forming method based on a plurality of image frames, image processing system, and digital camera | |
| JP2013508811A (en) | Blur image correction using spatial image prior probability | |
| WO2016139260A9 (en) | Method and system for real-time noise removal and image enhancement of high-dynamic range images | |
| EP1931134A1 (en) | Imaging device | |
| WO2021139635A1 (en) | Method and apparatus for generating super night scene image, and electronic device and storage medium | |
| JP2005039460A (en) | Image processing method and apparatus | |
| CN114519676A (en) | Bayer format-based raw image denoising device and method | |
| Adams Jr et al. | Digital camera image processing chain design | |
| US11915392B2 (en) | Image enhancement method and apparatus | |
| CN116894797A (en) | Compression denoising method and system for high-dynamic bayer image | |
| Deever et al. | Digital camera image formation: Processing and storage | |
| CN112241670B (en) | Image processing method and device | |
| CN114331893A (en) | Method, medium and electronic device for acquiring image noise | |
| EP3913572A1 (en) | Loss function for image reconstruction | |
| EP4058976A1 (en) | Device and method for pre-processing image data for a computer vision application | |
| CN117893455B (en) | Image brightness and contrast adjusting method | |
| KR101957850B1 (en) | Apparatus and method for reverse tone mapping using guided filter | |
| WO2024095624A1 (en) | Image processing device, learning method, and inference method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |