CN107945125B - Fuzzy image processing method integrating frequency spectrum estimation method and convolutional neural network - Google Patents
Fuzzy image processing method integrating frequency spectrum estimation method and convolutional neural network Download PDFInfo
- Publication number
- CN107945125B CN107945125B CN201711145578.6A CN201711145578A CN107945125B CN 107945125 B CN107945125 B CN 107945125B CN 201711145578 A CN201711145578 A CN 201711145578A CN 107945125 B CN107945125 B CN 107945125B
- Authority
- CN
- China
- Prior art keywords
- image
- value
- neural network
- convolutional neural
- blurred
- 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
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 20
- 238000001228 spectrum Methods 0.000 title claims abstract description 18
- 238000003672 processing method Methods 0.000 title claims abstract description 12
- 238000012545 processing Methods 0.000 claims abstract description 18
- 238000001914 filtration Methods 0.000 claims abstract description 14
- 230000033001 locomotion Effects 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 11
- 239000003086 colorant Substances 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 230000001902 propagating effect Effects 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 238000011161 development Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 abstract 1
- 230000000694 effects Effects 0.000 abstract 1
- 230000002708 enhancing effect Effects 0.000 abstract 1
- 230000006870 function Effects 0.000 description 12
- 230000000875 corresponding effect Effects 0.000 description 9
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
Images
Classifications
-
- G06T5/73—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4053—Super resolution, i.e. output image resolution higher than sensor resolution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/10—Image enhancement or restoration by non-spatial domain filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration by the use of local operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20172—Image enhancement details
- G06T2207/20201—Motion blur correction
Abstract
The invention provides a fuzzy image processing method integrating a frequency spectrum estimation method and a convolutional neural network, which comprises the steps of firstly carrying out graying processing on an input image, carrying out Fourier transform and generating a frequency spectrogram; secondly, calculating a fuzzy length and an angle by performing binarization processing on the frequency spectrogram and generating a horizontal projection diagram; and finally, restoring the blurred image by utilizing wiener filtering, and further enhancing the effect through a convolutional neural network. The method is simple and efficient, and has good development prospect.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a fuzzy image processing method fusing a frequency spectrum estimation method and a convolutional neural network.
Background
With the development of science and technology, the images are applied more and more frequently in daily life, and the images can be used in daily office work and online entertainment. In response to this, image restoration of degraded images is also becoming more and more important. Motion blurred images are one of the common types of blurred images. When we take pictures using a mobile phone, this often happens: at the moment we press the shutter, we shake our hands and then find that the picture is blurred. An image captured in this manner is referred to as a "motion-blurred image". As is known, the image restoration technique plays an important role in the whole image processing module, and its main purpose is to restore the blurred image to the original image quality standard. In image restoration, the motion-blurred image is an important part of the image and has practical significance, so that the method can be widely applied to real life and has a wide prospect.
Image restoration, which is an important part of image processing technology, is naturally receiving wide attention from scholars at home and abroad, and many related studies are being conducted. The following image restoration methods, from the initial deconvolution (i.e., inverse filtering) method to the subsequent linear restoration method, and the image blind deconvolution algorithm, are basically improved around the three methods. The main contents of the deconvolution restoration algorithm are as follows: power spectrum equalization, geometric mean filtering and Wiener filtering, etc., which are more traditional and very classical image restoration methods, are more suitable for the case that the linear space is not changed or the noise signal is not correlated. In the mid sixties, deconvolution processing of blurred images in the telescope due to atmospheric surge has been started using a point spread function (PSF for short) in Wiener filtering. The image blind deconvolution restoration method can directly estimate the true signal and the degradation function of the image according to the blurred image. However, the quality of the target image results obtained using this method is directly dependent on the choice of initial conditions, and the image results may not be unique. This method is not suitable if the image signal-to-noise ratio is low. The traditional Wiener filtering processing mode is to perform checking operation under the condition of knowing the angle and the length of the motion blur, which has great limitation on the practical use.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a fuzzy image processing method fusing a spectrum estimation method and a convolutional neural network, on the basis of traditional image restoration, the super-resolution realized by combining the spectrum estimation method and the convolutional neural network is combined to improve the image quality based on computer vision, and the use of the traditional Wiener filtering is converted into the fuzzy image which can be directly adapted to different motions through the change of point spread function parameters through a spectrogram analysis method.
In order to achieve the purpose, the technical scheme of the invention is as follows: a fuzzy image processing method fusing a spectrum estimation method and a convolutional neural network comprises the following steps:
step 1: inputting a blurred image;
step 2: carrying out graying processing on the blurred image, carrying out Fourier transform and generating a spectrogram;
and step 3: carrying out binarization processing on the spectrogram, generating a horizontal projection graph, and calculating a fuzzy length and an angle;
and 4, step 4: and restoring the blurred image by utilizing wiener filtering, and inputting the blurred image into a convolutional neural network to obtain a final image. Further, the step 2 specifically includes:
step 21: firstly, converting an image into YCbCr in a color space, then extracting a Y channel for gray processing, and adopting the following formula: gray (x, y) ═ α R (x, y) + β G (x, y) + γ B (x, y), where Gray (x, y) is the Gray value of the corresponding image position (x, y), R, G, B are the components of the three colors red, green, blue of the corresponding position, respectively, and α, β, γ are parameters;
step 22: performing one-dimensional Fourier transform on the grayscale image with N rows and N columns according to rows and columns by using the following formula:
firstly, performing discrete Fourier transform according to rows, then performing discrete Fourier transform according to columns, converting an image from a spatial domain F (x, y) into a frequency domain F (u, v), and finally obtaining a frequency domain value containing a real part and an imaginary part, wherein F (x, y) is a gray value of a corresponding position (x, y), u is a frequency component after row transform, v is a frequency component after column transform, and F (u, v) is a frequency spectrum value under corresponding u and v;
step 23: moving the origin of the spectrum image from the starting point (0,0) to the central point (N/2 ) of the image;
step 24: performing Fourier transform on complex values
Operating to obtain corresponding amplitude, wherein Re is a real part of a complex number, and Im is an imaginary part of the complex number;
step 25: and carrying out normalization operation on the amplitude map.
Further, α is 0.30, β is 0.59, and γ is 0.11.
Further, the step 3 specifically includes:
step 31: counting the number of pixels in each gray level in the spectrogram, calculating the proportion of the number of pixels in each gray level in the whole image, segmenting the images into a foreground and a background by utilizing a threshold value, and respectively calculating the probability w of dividing the images into the foreground0And its average gray value q0And probability of being divided into backgrounds w1And its average gray value q1Adopting a traversal method and using a formula sigma as w0*w1*(q0-q1)2Obtaining a segmentation threshold value which enables the sigma to be maximum, and then thresholding the image to obtain a non-black or white binary image;
step 32: dividing the binary image according to pixel points, searching according to rows from top to bottom, searching for the first row with white pixel points, searching for the first column with white pixel points from left to right, and overlapping the search results twice to obtain the target point A (x) at the top left corner1,y1) (ii) a Then, the same method is used to obtain the target point B (x) at the lower right corner2,y2) The following formula is adopted:
calculating to obtain a fuzzy angle theta of the motion blur;
step 33: clockwise rotating the binary image by an angle theta, calculating an accumulated value according to columns, obtaining a maximum value and a horizontal distance D of the image, then re-assigning half of the maximum value to the whole image which exceeds half of the maximum value, and traversing to obtain a minimum value region omega; the distance d to the first stripe of the central spot is calculated within Ω using the following equation:
the blur length L of the motion-blurred image is obtained.
Further, the step 4 specifically includes:
step 41: point spread function h of a sharp image f in motion blurL,θUnder the action ofThe blurred image g is obtained by adding the noise n, and the following equation is used: (h)L,θF) (x, y) + n (x, y) ═ g (x, y), deconvoluting the blurred image for image restoration;
step 42: inputting a series of training pictures { Xi,Yi}，XiFor the input original picture, YiFor the processed fuzzy picture, there are m groups of picture data in total, and mean square error is adopted
As a loss function, where Θ represents each parameter in the training process, and the F function is the function of Y under a series of parameters ΘiAnd (3) performing deblurring operation, adjusting parameters during training to minimize the mean square error, reversely propagating by using a random gradient algorithm, adjusting the parameters to minimize loss, and inputting the image subjected to wiener filtering into the trained convolutional neural network.
Compared with the prior art, the invention has the beneficial effects that:
the method comprises the steps of recovering a motion blurred image by combining a frequency spectrum estimation method and a convolution neural network, estimating the length and the angle of motion blur by utilizing a frequency spectrum image obtained after Fourier transform and combining a horizontal projection image, aiming at the fact that the traditional Wiener filtering processing mode is to carry out inspection operation under the condition that the angle and the length of the motion blur are known, and having great limitation on practical use.
Drawings
FIG. 1 is a flow chart of a fuzzy image processing method combining a spectrum estimation method and a convolutional neural network according to the present invention.
Detailed Description
The invention is further explained below with reference to the drawings and the embodiments.
Aiming at the condition that the traditional processing method is known in fuzzy angle and length, a method combining a point spread function and a convolution neural network is provided. The convolutional neural network based implicit learning characteristic from the master-slave data does not need to manually select a proper characteristic, and the training speed of the network can be increased and the complexity of the network can be reduced through operations such as weight sharing, maximum pooling and the like. The invention improves the image quality based on computer vision by combining the super-resolution realized by a deep convolutional neural network on the basis of the traditional image restoration.
As shown in fig. 1, the method for processing a blurred image by fusing a spectrum estimation method and a convolutional neural network provided by the present invention includes:
step 1: inputting a blurred image;
step 2: carrying out graying processing on the blurred image, carrying out Fourier transform and generating a spectrogram;
and step 3: carrying out binarization processing on the spectrogram, generating a horizontal projection graph, and calculating a fuzzy length and an angle;
and 4, step 4: and restoring the blurred image by utilizing wiener filtering, and inputting the blurred image into a convolutional neural network to obtain a final image. In this embodiment, the step 2 specifically includes:
step 21: firstly, converting an image into YCbCr in a color space, then extracting a Y channel for gray processing, and adopting the following formula: gray (x, y) ═ α R (x, y) + β G (x, y) + γ B (x, y), where Gray (x, y) is the Gray value of the corresponding image position (x, y), R, G, B are the components of the three colors red, green, blue of the corresponding position, respectively, and α, β, γ are parameters;
step 22: performing one-dimensional Fourier transform on the grayscale image with N rows and N columns according to rows and columns by using the following formula:
firstly, performing discrete Fourier transform according to rows, then performing discrete Fourier transform according to columns, converting an image from a spatial domain F (x, y) into a frequency domain F (u, v), and finally obtaining a frequency domain value containing a real part and an imaginary part, wherein F (x, y) is a gray value of a corresponding position (x, y), u is a frequency component after row transform, v is a frequency component after column transform, and F (u, v) isCorresponding to the spectral values under u and v;
step 23: moving the origin of the spectrum image from the starting point (0,0) to the central point (N/2 ) of the image;
step 24: performing Fourier transform on complex values
Operating to obtain corresponding amplitude, wherein Re is a real part of a complex number, and Im is an imaginary part of the complex number;
step 25: and carrying out normalization operation on the amplitude map.
In this embodiment, α is 0.30, β is 0.59, and γ is 0.11.
In this embodiment, the step 3 specifically includes:
step 31: counting the number of pixels in each gray level in the spectrogram, calculating the proportion of the number of pixels in each gray level in the whole image, segmenting the images into a foreground and a background by utilizing a threshold value, and respectively calculating the probability w of dividing the images into the foreground0And its average gray value q0And probability of being divided into backgrounds w1And its average gray value q1Adopting a traversal method and using a formula sigma as w0*w1*(q0-q1)2Obtaining a segmentation threshold value which enables the sigma to be maximum, and then thresholding the image to obtain a non-black or white binary image;
step 32: dividing the binary image according to pixel points, searching according to rows from top to bottom, searching for the first row with white pixel points, searching for the first column with white pixel points from left to right, and overlapping the search results twice to obtain the target point A (x) at the top left corner1,y1) (ii) a Then, the same method is used to obtain the target point B (x) at the lower right corner2,y2) The following formula is adopted:
calculating to obtain a fuzzy angle theta of the motion blur;
step 33: clockwise rotation angle theta is carried out on the binary image, the accumulated value is calculated according to columns, and the maximum value is obtainedAnd the horizontal distance D of the image, then assigning half of the maximum value again to the half of the maximum value in the whole image, and traversing to obtain a minimum value region omega; the distance d to the first stripe of the central spot is calculated within Ω using the following equation:
the blur length L of the motion-blurred image is obtained.
In this embodiment, the step 4 specifically includes:
step 41: point spread function h of a sharp image f in motion blurL,θWith the addition of noise n, the image becomes a blurred image g, using the following equation: (h)L,θF) (x, y) + n (x, y) ═ g (x, y), deconvoluting the blurred image for image restoration;
step 42: inputting a series of training pictures { Xi,Yi}，XiFor the input original picture, YiFor the processed fuzzy picture, there are m groups of picture data in total, and mean square error is adopted
As a loss function, where Θ represents each parameter in the training process, and the F function is the function of Y under a series of parameters ΘiAnd (3) performing deblurring operation, adjusting parameters during training to minimize the mean square error, reversely propagating by using a random gradient algorithm, adjusting the parameters to minimize loss, and inputting the image subjected to wiener filtering into the trained convolutional neural network.
The convolutional neural network can automatically learn from data without manually selecting proper characteristics, and the training speed of the network is increased and the complexity of the network is reduced through operations such as weight sharing, maximum pooling and the like.
Although the present invention has been described with reference to the preferred embodiments, it is not intended to limit the present invention, and those skilled in the art can make variations and modifications of the present invention without departing from the spirit and scope of the present invention by using the methods and technical contents disclosed above. The above description is only a preferred embodiment of the present invention, and all equivalent changes and modifications made in accordance with the claims of the present invention should be covered by the present invention.
Claims (4)
1. A fuzzy image processing method fusing a spectrum estimation method and a convolutional neural network is characterized by comprising the following steps:
step 1: inputting a blurred image;
step 2: carrying out graying processing on the blurred image, carrying out Fourier transform and generating a spectrogram;
and step 3: carrying out binarization processing on the spectrogram, generating a horizontal projection graph, and calculating a fuzzy length and an angle;
and 4, step 4: restoring the blurred image by utilizing wiener filtering, and inputting the blurred image into a convolutional neural network to obtain a final image; the step 3 specifically includes:
step 31: counting the number of pixels in each gray level in the spectrogram, calculating the proportion of the number of pixels in each gray level in the whole image, segmenting the images into a foreground and a background by utilizing a threshold value, and respectively calculating the probability w of dividing the images into the foreground0And its average gray value q0And probability of being divided into backgrounds w1And its average gray value q1Adopting a traversal method and using a formula sigma as w0*w1*(q0-q1)2Obtaining a segmentation threshold value which enables the sigma to be maximum, and then thresholding the image to obtain a non-black or white binary image;
step 32: dividing the binary image according to pixel points, searching according to rows from top to bottom, searching for the first row with white pixel points, searching for the first column with white pixel points from left to right, and overlapping the search results twice to obtain the target point A (x) at the top left corner1,y1) (ii) a Then, the same method is used to obtain the target point B (x) at the lower right corner2,y2) The following formula is adopted:
calculating to obtain a fuzzy angle theta of the motion blur;
step 33: clockwise rotating the binary image by an angle theta, calculating an accumulated value according to columns, obtaining a maximum value and a horizontal distance D of the image, then re-assigning half of the maximum value to the whole image which exceeds half of the maximum value, and traversing to obtain a minimum value region omega; the distance d to the first stripe of the central spot is calculated within Ω using the following equation:
the blur length L of the motion-blurred image is obtained.
2. The blurred image processing method according to claim 1, wherein the step 2 specifically comprises:
step 21: firstly, converting an image into YCbCr in a color space, then extracting a Y channel for gray processing, and adopting the following formula: gray (x, y) ═ α R (x, y) + β G (x, y) + γ B (x, y), where Gray (x, y) is the Gray value of the corresponding image position (x, y), R, G, B are the components of the three colors red, green, blue of the corresponding position, respectively, and α, β, γ are parameters;
step 22: performing one-dimensional Fourier transform on the grayscale image with N rows and N columns according to rows and columns by using the following formula:
firstly, performing discrete Fourier transform according to rows, then performing discrete Fourier transform according to columns, converting an image from a spatial domain F (x, y) into a frequency domain F (u, v), and finally obtaining a frequency domain value containing a real part and an imaginary part, wherein F (x, y) is a gray value of a corresponding position (x, y), u is a frequency component after row transform, v is a frequency component after column transform, and F (u, v) is a frequency spectrum value under corresponding u and v;
step 23: moving the origin of the spectrum image from the starting point (0,0) to the central point (N/2 ) of the image;
step 24: performing Fourier transform on complex values
Operating to obtain corresponding amplitude, wherein Re is a real part of a complex number, and Im is an imaginary part of the complex number;
step 25: and carrying out normalization operation on the amplitude map.
3. The blurred image processing method according to claim 2,
α＝0.30,β＝0.59,γ＝0.11。
4. the blurred image processing method according to claim 1, wherein the step 4 specifically comprises:
step 41: point spread function h of a sharp image f in motion blurL,θWith the addition of noise n, the image becomes a blurred image g, using the following equation: (h)L,θF) (x, y) + n (x, y) ═ g (x, y), deconvoluting the blurred image for image restoration;
step 42: inputting a series of training pictures { Xi,Yi}，XiFor the input original picture, YiFor the processed fuzzy picture, there are m groups of picture data in total, and mean square error is adopted
As a loss function, where Θ represents each parameter in the training process, and the F function is the function of Y under a series of parameters ΘiAnd (3) performing deblurring operation, adjusting parameters during training to minimize the mean square error, reversely propagating by using a random gradient algorithm, adjusting the parameters to minimize loss, and inputting the image subjected to wiener filtering into the trained convolutional neural network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711145578.6A CN107945125B (en) | 2017-11-17 | 2017-11-17 | Fuzzy image processing method integrating frequency spectrum estimation method and convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711145578.6A CN107945125B (en) | 2017-11-17 | 2017-11-17 | Fuzzy image processing method integrating frequency spectrum estimation method and convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107945125A CN107945125A (en) | 2018-04-20 |
| CN107945125B true CN107945125B (en) | 2021-06-22 |
Family
ID=61932816
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711145578.6A Active CN107945125B (en) | 2017-11-17 | 2017-11-17 | Fuzzy image processing method integrating frequency spectrum estimation method and convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107945125B (en) |
Families Citing this family (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10931853B2 (en) * | 2018-10-18 | 2021-02-23 | Sony Corporation | Enhanced color reproduction for upscaling |
| CN111105357B (en) * | 2018-10-25 | 2023-05-02 | 杭州海康威视数字技术股份有限公司 | Method and device for removing distortion of distorted image and electronic equipment |
| CN109284751A (en) * | 2018-10-31 | 2019-01-29 | 河南科技大学 | The non-textual filtering method of text location based on spectrum analysis and SVM |
| CN109410143B (en) * | 2018-10-31 | 2021-03-09 | 泰康保险集团股份有限公司 | Image enhancement method and device, electronic equipment and computer readable medium |
| CN110060220A (en) * | 2019-04-26 | 2019-07-26 | 中国科学院长春光学精密机械与物理研究所 | Based on the image de-noising method and system for improving BM3D algorithm |
| CN110264415B (en) * | 2019-05-24 | 2020-06-12 | 北京爱诺斯科技有限公司 | Image processing method for eliminating jitter blur |
| CN110443882B (en) * | 2019-07-05 | 2021-06-11 | 清华大学 | Light field microscopic three-dimensional reconstruction method and device based on deep learning algorithm |
| CN111080524A (en) * | 2019-12-19 | 2020-04-28 | 吉林农业大学 | Plant disease and insect pest identification method based on deep learning |
| CN111340724B (en) * | 2020-02-24 | 2021-02-19 | 卡莱特（深圳）云科技有限公司 | Image jitter removing method and device in LED screen correction process |
| CN111415313B (en) * | 2020-04-13 | 2022-08-30 | 展讯通信（上海）有限公司 | Image processing method, image processing device, electronic equipment and storage medium |
| CN111986102B (en) * | 2020-07-15 | 2024-02-27 | 万达信息股份有限公司 | Digital pathological image deblurring method |
| CN112712467B (en) * | 2021-01-11 | 2022-11-11 | 郑州科技学院 | Image processing method based on computer vision and color filter array |
| CN113807246A (en) * | 2021-09-16 | 2021-12-17 | 平安普惠企业管理有限公司 | Face recognition method, device, equipment and storage medium |
| CN116188254A (en) * | 2021-11-25 | 2023-05-30 | 北京字跳网络技术有限公司 | Fourier domain-based super-resolution image processing method, device, equipment and medium |
| CN114723642B (en) * | 2022-06-07 | 2022-08-19 | 深圳市资福医疗技术有限公司 | Image correction method and device and capsule endoscope |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101079149A (en) * | 2006-09-08 | 2007-11-28 | 浙江师范大学 | Noise-possessing movement fuzzy image restoration method based on radial basis nerve network |
| CN104655583A (en) * | 2015-02-04 | 2015-05-27 | 中国矿业大学 | Fourier-infrared-spectrum-based rapid coal quality recognition method |
| CN105825484A (en) * | 2016-03-23 | 2016-08-03 | 华南理工大学 | Depth image denoising and enhancing method based on deep learning |
-
2017
- 2017-11-17 CN CN201711145578.6A patent/CN107945125B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101079149A (en) * | 2006-09-08 | 2007-11-28 | 浙江师范大学 | Noise-possessing movement fuzzy image restoration method based on radial basis nerve network |
| CN104655583A (en) * | 2015-02-04 | 2015-05-27 | 中国矿业大学 | Fourier-infrared-spectrum-based rapid coal quality recognition method |
| CN105825484A (en) * | 2016-03-23 | 2016-08-03 | 华南理工大学 | Depth image denoising and enhancing method based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| Blurred image restoration：A fast method of finding the motion length and angle;Michal Dobes等;《Digtial Signal Processing》;20100327;全文 * |
| 运动模糊车牌识别关键技术研究;史海玲;《中国优秀硕士学位论文全文数据库 工程科技II辑》;20160615;论文第1-2章 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107945125A (en) | 2018-04-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107945125B (en) | Fuzzy image processing method integrating frequency spectrum estimation method and convolutional neural network | |
| Claus et al. | Videnn: Deep blind video denoising | |
| Li et al. | Edge-preserving decomposition-based single image haze removal | |
| CN108921800B (en) | Non-local mean denoising method based on shape self-adaptive search window | |
| WO2016206087A1 (en) | Low-illumination image processing method and device | |
| US20180122051A1 (en) | Method and device for image haze removal | |
| CN110136055B (en) | Super resolution method and device for image, storage medium and electronic device | |
| CN108932699B (en) | Three-dimensional matching harmonic filtering image denoising method based on transform domain | |
| WO2014070273A1 (en) | Recursive conditional means image denoising | |
| Al-Hatmi et al. | A review of Image Enhancement Systems and a case study of Salt &pepper noise removing | |
| CN111445424A (en) | Image processing method, image processing device, mobile terminal video processing method, mobile terminal video processing device, mobile terminal video processing equipment and mobile terminal video processing medium | |
| CN105719251B (en) | A kind of compression degraded image restored method that Linear Fuzzy is moved for big picture | |
| CN107256539B (en) | Image sharpening method based on local contrast | |
| CN111353955A (en) | Image processing method, device, equipment and storage medium | |
| Das et al. | A comparative study of single image fog removal methods | |
| Patil et al. | Bilateral filter for image denoising | |
| CN107945119B (en) | Method for estimating correlated noise in image based on Bayer pattern | |
| CN111415317B (en) | Image processing method and device, electronic equipment and computer readable storage medium | |
| CN110852947B (en) | Infrared image super-resolution method based on edge sharpening | |
| CN108573478B (en) | Median filtering method and device | |
| Kaur et al. | An improved adaptive bilateral filter to remove gaussian noise from color images | |
| CN115829967A (en) | Industrial metal surface defect image denoising and enhancing method | |
| CN110648291B (en) | Unmanned aerial vehicle motion blurred image restoration method based on deep learning | |
| Sophia et al. | An efficient method for Blind Image Restoration using GAN | |
| Zhang et al. | Uav remote sensing image dehazing based on saliency guided two-scaletransmission correction |
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 |