CN110264415B - Image processing method for eliminating jitter blur - Google Patents
Image processing method for eliminating jitter blur Download PDFInfo
- Publication number
- CN110264415B CN110264415B CN201910442125.2A CN201910442125A CN110264415B CN 110264415 B CN110264415 B CN 110264415B CN 201910442125 A CN201910442125 A CN 201910442125A CN 110264415 B CN110264415 B CN 110264415B
- Authority
- CN
- China
- Prior art keywords
- image
- fuzzy
- processing
- texture
- distribution
- 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.)
- Active
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 22
- 230000009467 reduction Effects 0.000 claims abstract description 19
- 230000009466 transformation Effects 0.000 claims abstract description 18
- 238000013135 deep learning Methods 0.000 claims description 36
- 238000000034 method Methods 0.000 claims description 25
- 238000006243 chemical reaction Methods 0.000 claims description 24
- 230000008569 process Effects 0.000 claims description 19
- 238000001914 filtration Methods 0.000 claims description 15
- 238000001228 spectrum Methods 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 9
- 230000003044 adaptive effect Effects 0.000 claims description 6
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 1
Images
Classifications
-
- G06T5/70—
-
- G06T5/73—
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/80—Camera processing pipelines; Components thereof
Abstract
The invention provides an image processing method for eliminating jitter and blur, which comprises the steps of sequentially carrying out noise reduction processing and sharpening processing on an image, determining blur information corresponding to the image based on the result of the sharpening processing, carrying out first image operation processing on the image through a depth learning mode based on the blur information so as to obtain an intermediate transformation image, and finally carrying out second image operation processing on the intermediate transformation image through the depth learning mode so as to eliminate the jitter and blur in the image so as to realize the sharpening processing on the image.
Description
Technical Field
The present invention relates to the field of image processing technologies, and in particular, to an image processing method for eliminating blur.
Background
With the popularization of smart phones, smart phones have been integrated into life and work of people, and people can realize different forms of operations such as instant messaging, web browsing, entertainment, light office and the like through smart phones. The smart phone can be developed greatly, and the smart phone combines a traditional mobile phone with different functional modules such as a camera, so that the smart phone is not limited to the traditional call function any more, and can also perform other intelligent operations formed in different ways. The camera function of the camera in the smart phone is continuously enhanced, so that people can shoot images anytime and anywhere and share the shot images through the cloud in time.
Although the camera of the smart phone has a powerful image shooting function, the image shooting is always completed by manual operation, which causes shaking of the shot hand due to the influence of external environment or the self factor of the photographer during the shooting process, so that the camera is out of focus, and accordingly the shot image is necessarily blurred. Since the situation of hand shake is widely existed and unpredictable and artificially avoided, there is an uncertain factor of shake blur all the time during image capturing. In order to avoid blurring of a shot image caused by hand shake, the prior art arranges an anti-shake lens unit in an imaging lens system of a camera, the anti-shake lens unit can move under the driving of a control signal, the movement of the anti-shake lens unit can effectively counteract the out-of-focus condition caused by hand shake, although the image blurring caused by hand shake can be effectively reduced by arranging an anti-shake component in a hardware form in the camera, the volume and the hardware cost of the camera are increased by the mode, and meanwhile, higher requirements are provided for the adjustment of the control signal, which obviously has a conflict place for a smart phone seeking miniaturization, and more rigorous requirements are provided for the control performance of the smart phone. Therefore, in the prior art, the processing mode of the image blur caused by hand shake is realized in a hardware mode, and at present, no technology for overcoming the image blur caused by hand shake is realized in a software processing mode.
Disclosure of Invention
The image processing method for eliminating the shaking blur comprises the steps of sequentially carrying out noise reduction processing and sharpening processing on an image, determining the corresponding blurring information of the image based on the result of the sharpening processing, carrying out first image operation processing on the image according to an image spectrum through a deep learning mode based on the blurring information to obtain an intermediate conversion image, and finally carrying out second image operation processing on the intermediate conversion image according to the image resolution through the deep learning mode to eliminate the shaking blur in the image so as to realize the sharpening processing on the image. The image processing method for eliminating the shaking blur is different from a mode depending on hardware equipment in the prior art for reducing the image blur caused by hand shaking, the image processing method adopts a form of subsequent calculation processing on a shot image, carries out analysis processing on the image on an algorithm level, does not need the support of any additional hardware equipment, and only needs to carry out corresponding noise reduction processing and sharpening processing on the image after the image shooting is finished so as to obtain corresponding image blur information, and then carries out corresponding neural network deep learning processing on the image based on the image blur information, so that the blurred pixels in the image are eliminated, and finally the image is clarified.
The invention provides an image processing method for eliminating shake blur, which is characterized by comprising the following steps:
step (1), sequentially carrying out noise reduction processing and sharpening processing on an image;
step (2), based on the result of the sharpening process, determining fuzzy information corresponding to the image;
step (3) of performing first image operation processing on the image through a deep learning mode based on the fuzzy information to obtain an intermediate transformation image;
step (4), through a deep learning mode, second image operation processing related to image resolution is carried out on the intermediate conversion image, and shaking blurring in the image is eliminated, so that the image is sharpened;
further, in the step (1), the performing noise reduction processing on the image specifically includes,
step (A101), acquiring brightness distribution information of the image, and dividing the image into a plurality of different texture areas with different image texture distribution states based on the brightness distribution information;
step (A102), calculating and obtaining a brightness component and/or a texture component corresponding to each texture region in the plurality of different texture regions, and dividing all pixel points in the image into first type pixel points and second type pixel points according to the brightness component and/or the texture component;
step (A103), respectively executing adaptive filtering and noise reduction processing on the first type pixel points and the second type pixel points;
further, in the step (a101), acquiring brightness distribution information of the image, dividing the image into a plurality of different texture regions having different image texture distribution states based on the brightness distribution information specifically includes,
a step (a1011) of acquiring luminance distribution information and hue distribution information of the image, and calculating luminance-hue correlation information corresponding to the image in the entire image area based on the luminance distribution information and the hue distribution information;
a step (A1012) of determining an image texture distribution state of the image corresponding to the whole image area according to the brightness-tone correlation information corresponding to the whole image area, wherein the image texture distribution state at least comprises light and dark stripe distribution information about the image and/or three primary color distribution information about the image;
a step (a1013) of dividing the image into a plurality of different texture regions according to the image texture distribution state;
further, in the step (a102), dividing all the pixels in the image into first type pixels and second type pixels according to the luminance component and/or the texture component specifically includes,
step (A1021), according to the brightness component and the texture component, calculating a brightness-texture interference coefficient corresponding to each pixel point in the image, and comparing the brightness-texture interference coefficient with a preset interference coefficient range;
step (a1022), if the brightness-texture interference coefficient is within the preset interference coefficient range, determining that the corresponding pixel point is a first type pixel point, and if the brightness-texture interference coefficient is not within the preset interference coefficient range, determining that the corresponding pixel segment is a second type pixel point, where the first type pixel point is a noise type pixel point and the second type pixel point is a non-noise type pixel point;
alternatively, the first and second electrodes may be,
in the step (a103), the performing adaptive filtering and denoising on the first type pixels and the second type pixels specifically includes,
step (A1031), performing Gaussian filtering processing or Kalman filtering processing on a full pixel region in the first type pixel point so as to enable a noise coefficient of the first type pixel point to accord with a preset condition;
step (A1032), smoothing filtering processing is carried out on the high-frequency pixel region in the second type pixel point, so that the noise coefficient of the second type pixel point meets the preset condition;
further, in the step (1), the sharpening process on the image specifically includes,
step (B101), performing image fuzzification conversion on the noise-reduced image obtained after the noise reduction treatment through a Sobel operator;
step (B102), determining an edge region of the blurred image after the image blurring conversion, wherein the edge region at least comprises a boundary of the blurred image and a region corresponding to a preset distance range inside and outside the boundary;
step (B103) of obtaining a fuzzy feature corresponding to the edge region, and calculating to obtain a fuzzy kernel corresponding to the edge region, wherein the fuzzy feature includes at least one of a fuzzy direction, a fuzzy track and a fuzzy angle;
further, in the step (2), based on the result of the sharpening process, determining that the blur information corresponding to the image specifically includes,
a step (201) of determining a fuzzy association coefficient of the whole image area based on fuzzy kernel information about the image obtained by the sharpening process;
step (202), dividing a plurality of regions with different fuzzy degrees on the whole region of the image based on the distribution state of the fuzzy association coefficient of the whole region of the image;
step (203), calculating the pixel fuzzy distribution trend information corresponding to each fuzzy degree area based on the division results of the plurality of different fuzzy degree areas, and taking the pixel fuzzy distribution trend information as the fuzzy information;
further, the step (3) of performing a first image operation process on the image spectrum through a deep learning mode based on the blur information to obtain an intermediate transformed image specifically includes,
a step (301) of performing pixel fitting transformation processing on the image based on at least one of a fuzzy vector, a fuzzy direction, a fuzzy track and a fuzzy angle corresponding to the pixels of different areas of the image in the fuzzy information, so as to obtain a corresponding pixel fitting image;
step (302), a first image operation processing related to an image frequency spectrum is carried out on the pixel fitting image through a deep learning mode, and the intermediate conversion image is obtained through calculation;
further, the step (302) of performing a first image operation process related to an image spectrum on the pixel-fitted image in a deep learning mode and calculating the intermediate transformed image specifically includes,
performing image deconvolution operation corresponding to a least square method related to an image frequency spectrum and a Landweber iteration method on the pixel fitting image through a deep learning mode, so as to calculate and obtain the intermediate transformation image;
further, in the step (4), performing second image operation processing with respect to an image resolution on the intermediate transformed image by a mode of deep learning, the removing of the blur in the image specifically includes,
step (401), acquiring a resolution distribution state of the intermediate conversion image in the whole area, and judging whether the resolution distribution state meets a preset resolution distribution condition;
step (402), if the resolution distribution state meets the preset resolution distribution condition, performing the second image operation processing on the intermediate conversion image, and if the resolution distribution state does not meet the preset resolution distribution condition, performing fuzzy pixel point elimination processing on the intermediate conversion image until the resolution distribution state meets the preset resolution distribution condition;
step (403), executing the second image operation processing on the intermediate transformation image which accords with the preset respective rate distribution condition through a deep learning mode so as to eliminate jitter and blur in the image;
further, in the step (403), the performing the second image operation process on the intermediate transformed image that meets the preset distribution condition specifically includes,
and executing calculation processing of super-resolution restoration on all pixels of the intermediate transformation image through a deep learning mode, so as to realize the improved restoration of all pixels in at least one of brightness values, chroma values and hue values, so as to realize the sharpening processing and eliminate the shake blur in the image.
Compared with the prior art, the image processing method for eliminating the shake blur sequentially carries out noise reduction processing and sharpening processing on an image, determines the blur information corresponding to the image based on the result of the sharpening processing, carries out first image operation processing on the image through a deep learning mode based on the blur information so as to obtain an intermediate transformation image, and finally carries out second image operation processing on the intermediate transformation image through the deep learning mode so as to eliminate the shake blur in the image, so that the image is sharpened. The image processing method for eliminating the shaking blur is different from a mode depending on hardware equipment in the prior art for reducing the image blur caused by hand shaking, the image processing method adopts a form of subsequent calculation processing on a shot image, carries out analysis processing on the image on an algorithm level, does not need the support of any additional hardware equipment, and only needs to carry out corresponding noise reduction processing and sharpening processing on the image after the image shooting is finished so as to obtain corresponding image blur information, and then carries out corresponding neural network deep learning processing on the image based on the image blur information, so that the blurred pixels in the image are eliminated, and finally the image is clarified.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic flow chart of an image processing method for removing blur.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Fig. 1 is a flowchart illustrating an image processing method for removing blur according to an embodiment of the present invention. The image processing method for eliminating the shake blur comprises the following steps:
and (1) sequentially carrying out noise reduction processing and sharpening processing on the image.
Preferably, in the step (1), the noise reduction processing of the image specifically includes,
step (A101), obtaining the brightness distribution information of the image, and dividing the image into a plurality of different texture areas with different image texture distribution states based on the brightness distribution information;
step (A102), calculating and obtaining a brightness component and/or a texture component corresponding to each texture region in the plurality of different texture regions, and dividing all pixel points in the image into a first type pixel point and a second type pixel point according to the brightness component and/or the texture component;
and (A103) performing adaptive filtering and noise reduction processing on the first type pixel point and the second type pixel point respectively.
Preferably, in the step (a101), acquiring brightness distribution information of the image, dividing the image into a plurality of different texture regions having different texture distribution states of the image based on the brightness distribution information specifically includes,
a step (a1011) of acquiring luminance distribution information and hue distribution information of the image, and calculating luminance-hue correlation information corresponding to the image in the entire image area based on the luminance distribution information and the hue distribution information;
step (A1012), according to the brightness-tone correlation information corresponding to the whole image area, determining an image texture distribution state corresponding to the whole image area, wherein the image texture distribution state at least comprises light and shade stripe distribution information about the image and/or three primary color distribution information about the image;
and (A1013) dividing the image into a plurality of different texture areas according to the texture distribution state of the image.
Preferably, in the step (a102), dividing all the pixels in the image into the first type pixels and the second type pixels according to the luminance component and/or the texture component specifically includes,
step (A1021), according to the brightness component and the texture component, calculating a brightness-texture interference coefficient corresponding to each pixel point in the image, and comparing the brightness-texture interference coefficient with a preset interference coefficient range;
step (a1022), if the luminance-texture interference coefficient is within the preset interference coefficient range, determining the corresponding pixel point as a first type pixel point, and if the luminance-texture interference coefficient is not within the preset interference coefficient range, determining the corresponding pixel segment as a second type pixel point, where the first type pixel point is a noise type pixel point and the second type pixel point is a non-noise type pixel point.
Preferably, in the step (a103), the performing adaptive filtering and denoising on the first type pixels and the second type pixels specifically includes,
step (A1031), performing Gaussian filtering processing or Kalman filtering processing on a full pixel region in the first type pixel point so as to enable a noise coefficient of the first type pixel point to accord with a preset condition;
and (A1032), performing smoothing filtering processing on the high-frequency pixel region in the second type pixel point so as to enable the noise coefficient of the second type pixel point to meet a preset condition.
Preferably, in the step (1), the sharpening process on the image specifically includes,
step (B101), the noise-reduced image obtained after the noise reduction processing is subjected to image fuzzification conversion through a Sobel operator;
step (B102), determining an edge area of the blurred image after the image blurring conversion, wherein the edge area at least comprises a boundary of the blurred image and an area corresponding to a preset distance range inside and outside the boundary;
step (B103) of obtaining a blur feature corresponding to the edge region, and calculating a blur kernel corresponding to the edge region, where the blur feature includes at least one of a blur direction, a blur trajectory, and a blur angle.
And (2) determining fuzzy information corresponding to the image based on the result of the sharpening processing.
Preferably, in the step (2), based on the result of the sharpening process, determining that the blur information corresponding to the image specifically includes,
step (201), based on the fuzzy kernel information about the image obtained by the sharpening process, determining a fuzzy association coefficient of the whole area of the image;
step (202), based on the distribution state of the fuzzy correlation coefficient of the whole area of the image, dividing the whole area of the image into a plurality of areas with different fuzzy degrees;
and (203) calculating the pixel fuzzy distribution trend information corresponding to each fuzzy degree area based on the division results of the plurality of different fuzzy degree areas, and taking the pixel fuzzy distribution trend information as the fuzzy information.
And (3) performing first image operation processing on the image spectrum through a deep learning mode based on the fuzzy information to obtain an intermediate conversion image.
Preferably, in the step (3), the performing a first image operation process on the image spectrum through a deep learning mode based on the blur information to obtain an intermediate transformed image specifically includes,
a step (301) of performing pixel fitting transformation processing on the image based on at least one of a blur vector, a blur direction, a blur trajectory and a blur angle corresponding to pixels in different areas of the image of the blur information, thereby obtaining a corresponding pixel-fitted image;
and (302) performing first image operation processing related to an image spectrum on the pixel fitting image in a deep learning mode, and calculating to obtain the intermediate conversion image.
Preferably, in the step (302), performing a first image operation process related to an image spectrum on the pixel-fitting image through a deep learning mode, and calculating the intermediate transformed image specifically includes,
and performing image deconvolution operation corresponding to a least square method related to an image frequency spectrum and a Landweber iteration method on the pixel fitting image through a deep learning mode, so as to calculate and obtain the intermediate transformation image.
And (4) performing second image operation processing on the intermediate conversion image according to the image resolution through a deep learning mode, and eliminating the shake blur in the image so as to realize the sharpening processing on the image.
Preferably, in the step (4), the performing of the second image operation processing with respect to the image resolution on the intermediate transformed image by the mode of the deep learning, the removing of the shake blur in the image specifically includes,
step (401), obtaining the resolution distribution state of the intermediate conversion image in the whole area, and judging whether the resolution distribution state meets the preset resolution distribution condition;
step (402), if the resolution distribution state conforms to the preset resolution distribution condition, performing the second image operation processing on the intermediate conversion image, and if the resolution distribution state does not conform to the preset resolution distribution condition, performing fuzzy pixel point elimination processing on the intermediate conversion image until the resolution distribution state conforms to the preset resolution distribution condition;
and (403) executing the second image operation processing on the intermediate transformed image meeting the preset respective rate distribution condition through a deep learning mode to eliminate jitter and blur in the image.
Preferably, in the step (403), the performing the second image operation process on the intermediate transformed image that meets the preset respective rate distribution condition specifically includes,
and performing super-resolution restoration calculation processing on all pixels of the intermediate transformation image through a deep learning mode, so as to realize the promotion and restoration of all pixels in at least one of a brightness value, a chromatic value and a hue value, so as to realize the sharpening processing and eliminate the shake blur in the image.
As can be seen from the foregoing embodiments, the image processing method for removing blur and blur sequentially performs noise reduction processing and sharpening processing on an image, determines blur information corresponding to the image based on a result of the sharpening processing, performs first image operation processing on an image spectrum on the image in a deep learning mode based on the blur information to obtain an intermediate transformed image, and performs second image operation processing on the intermediate transformed image in an image resolution in the deep learning mode to remove blur and blur in the image, thereby implementing sharpening processing on the image. The image processing method for eliminating the shaking blur is different from a mode depending on hardware equipment in the prior art for reducing the image blur caused by hand shaking, the image processing method adopts a form of subsequent calculation processing on a shot image, carries out analysis processing on the image on an algorithm level, does not need the support of any additional hardware equipment, and only needs to carry out corresponding noise reduction processing and sharpening processing on the image after the image shooting is finished so as to obtain corresponding image blur information, and then carries out corresponding neural network deep learning processing on the image based on the image blur information, so that the blurred pixels in the image are eliminated, and finally the image is clarified.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (4)
1. An image processing method for removing a shake blur, the image processing method for removing a shake blur comprising the steps of:
step (1), sequentially carrying out noise reduction processing and sharpening processing on an image;
step (2), based on the result of the sharpening process, determining fuzzy information corresponding to the image;
step (3) of performing first image operation processing on the image through a deep learning mode based on the fuzzy information to obtain an intermediate transformation image;
step (4), through a deep learning mode, second image operation processing related to image resolution is carried out on the intermediate conversion image, and shaking blurring in the image is eliminated, so that the image is sharpened;
in the step (1), the performing noise reduction processing on the image specifically includes,
step (A101), acquiring brightness distribution information of the image, and dividing the image into a plurality of different texture areas with different image texture distribution states based on the brightness distribution information;
step (A102), calculating and obtaining a brightness component and/or a texture component corresponding to each texture region in the plurality of different texture regions, and dividing all pixel points in the image into first type pixel points and second type pixel points according to the brightness component and/or the texture component;
step (A103), respectively executing adaptive filtering and noise reduction processing on the first type pixel points and the second type pixel points;
in the step (A102), dividing all the pixels in the image into first-type pixels and second-type pixels according to the luminance component and/or the texture component,
step (A1021), according to the brightness component and the texture component, calculating a brightness-texture interference coefficient corresponding to each pixel point in the image, and comparing the brightness-texture interference coefficient with a preset interference coefficient range;
step (a1022), if the brightness-texture interference coefficient is within the preset interference coefficient range, determining that the corresponding pixel point is a first type pixel point, and if the brightness-texture interference coefficient is not within the preset interference coefficient range, determining that the corresponding pixel segment is a second type pixel point, where the first type pixel point is a noise type pixel point and the second type pixel point is a non-noise type pixel point;
alternatively, the first and second electrodes may be,
in the step (a103), the performing adaptive filtering and denoising on the first type pixels and the second type pixels specifically includes,
step (A1031), performing Gaussian filtering processing or Kalman filtering processing on a full pixel region in the first type pixel point so as to enable a noise coefficient of the first type pixel point to accord with a preset condition;
step (A1032), smoothing filtering processing is carried out on the high-frequency pixel region in the second type pixel point, so that the noise coefficient of the second type pixel point meets the preset condition;
in the step (3), the obtaining of the intermediate transformed image by performing the first image operation processing on the image spectrum through the mode of the deep learning based on the blur information specifically includes,
a step (301) of performing pixel fitting transformation processing on the image based on at least one of a fuzzy vector, a fuzzy direction, a fuzzy track and a fuzzy angle corresponding to the pixels of different areas of the image in the fuzzy information, so as to obtain a corresponding pixel fitting image;
step (302), a first image operation processing related to an image frequency spectrum is carried out on the pixel fitting image through a deep learning mode, and the intermediate conversion image is obtained through calculation;
in the step (302), the first image operation processing related to the image spectrum is performed on the pixel-fitting image in a deep learning mode, and the calculation to obtain the intermediate transformation image specifically includes,
performing image deconvolution operation corresponding to a least square method related to an image frequency spectrum and a Landweber iteration method on the pixel fitting image through a deep learning mode, so as to calculate and obtain the intermediate transformation image;
in the step (4), performing second image operation processing with respect to an image resolution on the intermediate transformed image in a mode of deep learning, the removing of the blur in the image specifically includes,
step (401), acquiring a resolution distribution state of the intermediate conversion image in the whole area, and judging whether the resolution distribution state meets a preset resolution distribution condition;
step (402), if the resolution distribution state meets the preset resolution distribution condition, performing the second image operation processing on the intermediate conversion image, and if the resolution distribution state does not meet the preset resolution distribution condition, performing fuzzy pixel point elimination processing on the intermediate conversion image until the resolution distribution state meets the preset resolution distribution condition;
step (403), executing the second image operation processing on the intermediate transformation image which accords with the preset respective rate distribution condition through a deep learning mode so as to eliminate jitter and blur in the image;
in the step (403), the performing the second image operation process on the intermediate transformed image that meets the preset respective rate distribution condition specifically includes,
and executing calculation processing of super-resolution restoration on all pixels of the intermediate transformation image through a deep learning mode, so as to realize the improved restoration of all pixels in at least one of brightness values, chroma values and hue values, so as to realize the sharpening processing and eliminate the shake blur in the image.
2. The image processing method of removing shake blur according to claim 1, characterized in that:
in the step (a101), acquiring brightness distribution information of the image, dividing the image into a plurality of different texture regions having different image texture distribution states based on the brightness distribution information specifically includes,
a step (a1011) of acquiring luminance distribution information and hue distribution information of the image, and calculating luminance-hue correlation information corresponding to the image in the entire image area based on the luminance distribution information and the hue distribution information;
a step (A1012) of determining an image texture distribution state of the image corresponding to the whole image area according to the brightness-tone correlation information corresponding to the whole image area, wherein the image texture distribution state at least comprises light and dark stripe distribution information about the image and/or three primary color distribution information about the image;
and (A1013) dividing the image into a plurality of different texture areas according to the image texture distribution state.
3. The image processing method of removing shake blur according to claim 1, characterized in that:
in the step (1), the sharpening process on the image specifically includes,
step (B101), performing image fuzzification conversion on the noise-reduced image obtained after the noise reduction treatment through a Sobel operator;
step (B102), determining an edge region of the blurred image after the image blurring conversion, wherein the edge region at least comprises a boundary of the blurred image and a region corresponding to a preset distance range inside and outside the boundary;
and (B103) acquiring a fuzzy feature corresponding to the edge region, and calculating to obtain a fuzzy kernel corresponding to the edge region, wherein the fuzzy feature comprises at least one of a fuzzy direction, a fuzzy track and a fuzzy angle.
4. The image processing method of removing shake blur according to claim 1, characterized in that:
in the step (2), determining that the blur information corresponding to the image specifically includes, based on a result of the sharpening process,
a step (201) of determining a fuzzy association coefficient of the whole image area based on fuzzy kernel information about the image obtained by the sharpening process;
step (202), dividing a plurality of regions with different fuzzy degrees on the whole region of the image based on the distribution state of the fuzzy association coefficient of the whole region of the image;
and (203) calculating the pixel fuzzy distribution trend information corresponding to each fuzzy degree area based on the division results of the plurality of different fuzzy degree areas, and taking the pixel fuzzy distribution trend information as the fuzzy information.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910442125.2A CN110264415B (en) | 2019-05-24 | 2019-05-24 | Image processing method for eliminating jitter blur |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910442125.2A CN110264415B (en) | 2019-05-24 | 2019-05-24 | Image processing method for eliminating jitter blur |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110264415A CN110264415A (en) | 2019-09-20 |
CN110264415B true CN110264415B (en) | 2020-06-12 |
Family
ID=67915390
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910442125.2A Active CN110264415B (en) | 2019-05-24 | 2019-05-24 | Image processing method for eliminating jitter blur |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110264415B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111340735B (en) * | 2020-03-04 | 2021-04-23 | 卡莱特云科技股份有限公司 | LED screen body correction method, device and terminal |
CN112019749B (en) * | 2020-09-02 | 2021-06-25 | 中国银行股份有限公司 | Camera adjusting method and device |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106530256A (en) * | 2016-11-18 | 2017-03-22 | 四川长虹电器股份有限公司 | Improved-deep-learning-based intelligent camera image blind super-resolution system |
CN107730453A (en) * | 2017-11-13 | 2018-02-23 | 携程计算机技术(上海)有限公司 | Picture quality method for improving |
CN109145796A (en) * | 2018-08-13 | 2019-01-04 | 福建和盛高科技产业有限公司 | A kind of identification of electric power piping lane fire source and fire point distance measuring method based on video image convergence analysis algorithm |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1889223A1 (en) * | 2005-06-07 | 2008-02-20 | Thomson Licensing | Content-based gaussian noise reduction for still image, video and film |
CN102184533B (en) * | 2011-06-10 | 2012-10-24 | 西安电子科技大学 | Non-local-restriction-based total variation image deblurring method |
CN104616257A (en) * | 2015-01-26 | 2015-05-13 | 山东省计算中心(国家超级计算济南中心) | Recovery evidence obtaining method for blurred degraded digital images in administration of justice |
CN106447626B (en) * | 2016-09-07 | 2019-06-07 | 华中科技大学 | A kind of fuzzy core size estimation method and system based on deep learning |
CN107945125B (en) * | 2017-11-17 | 2021-06-22 | 福州大学 | Fuzzy image processing method integrating frequency spectrum estimation method and convolutional neural network |
CN108389163A (en) * | 2018-01-22 | 2018-08-10 | 西安工程大学 | A kind of self-adapting enhancement method based on half-light coloured image |
CN108765511B (en) * | 2018-05-30 | 2023-03-24 | 重庆大学 | Ultrasonic image super-resolution reconstruction method based on deep learning |
CN109377459B (en) * | 2018-09-30 | 2022-03-18 | 国网智能科技股份有限公司 | Super-resolution deblurring method of generative confrontation network |
-
2019
- 2019-05-24 CN CN201910442125.2A patent/CN110264415B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106530256A (en) * | 2016-11-18 | 2017-03-22 | 四川长虹电器股份有限公司 | Improved-deep-learning-based intelligent camera image blind super-resolution system |
CN107730453A (en) * | 2017-11-13 | 2018-02-23 | 携程计算机技术(上海)有限公司 | Picture quality method for improving |
CN109145796A (en) * | 2018-08-13 | 2019-01-04 | 福建和盛高科技产业有限公司 | A kind of identification of electric power piping lane fire source and fire point distance measuring method based on video image convergence analysis algorithm |
Also Published As
Publication number | Publication date |
---|---|
CN110264415A (en) | 2019-09-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107945125B (en) | Fuzzy image processing method integrating frequency spectrum estimation method and convolutional neural network | |
KR101313911B1 (en) | Method and apparatus for processing an image | |
KR100791391B1 (en) | Method and apparatus for noise reduction | |
WO2015184408A1 (en) | Scene motion correction in fused image systems | |
CN107248148B (en) | Image noise reduction method and system | |
CN109743473A (en) | Video image 3 D noise-reduction method, computer installation and computer readable storage medium | |
CN109816608B (en) | Low-illumination image self-adaptive brightness enhancement method based on noise suppression | |
CN107492077B (en) | Image deblurring method based on self-adaptive multidirectional total variation | |
CN110264415B (en) | Image processing method for eliminating jitter blur | |
WO2023273868A1 (en) | Image denoising method and apparatus, terminal, and storage medium | |
CN109214996B (en) | Image processing method and device | |
CN112583999B (en) | Method for detecting lens dirt of camera module | |
Mangiat et al. | Spatially adaptive filtering for registration artifact removal in HDR video | |
US9014503B2 (en) | Noise-reduction method and apparatus | |
KR20070078463A (en) | Apparatus and method for reducing noise from image sensor | |
CN113962898A (en) | Low-illumination image enhancement method based on illumination map optimization and adaptive gamma correction | |
CN110136085B (en) | Image noise reduction method and device | |
KR101101434B1 (en) | Apparatus for improving sharpness of image | |
KR101089902B1 (en) | Apparatus and method for estimating edge of digital image | |
CN108573478B (en) | Median filtering method and device | |
CN111091506A (en) | Image processing method and device, storage medium and electronic equipment | |
Fursov et al. | Two-step technology for improving details of images captured with mobile devices | |
Sethi et al. | Local enhancement of SLIC segmented underwater images using gray world based algorithm | |
CN114119411A (en) | Fog noise video image recovery method, device, equipment and medium | |
CN110852947B (en) | Infrared image super-resolution method based on edge sharpening |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |