CN105376462B - A kind of content benefit of Polluted area in video paints method - Google Patents
A kind of content benefit of Polluted area in video paints method Download PDFInfo
- Publication number
- CN105376462B CN105376462B CN201510760914.2A CN201510760914A CN105376462B CN 105376462 B CN105376462 B CN 105376462B CN 201510760914 A CN201510760914 A CN 201510760914A CN 105376462 B CN105376462 B CN 105376462B
- Authority
- CN
- China
- Prior art keywords
- current frame
- frame image
- matrix
- polluted area
- 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.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
- H04N5/144—Movement detection
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/60—Noise processing, e.g. detecting, correcting, reducing or removing noise
- H04N25/63—Noise processing, e.g. detecting, correcting, reducing or removing noise applied to dark current
Abstract
Content benefit the invention discloses the Polluted area in a kind of video paints method, comprises the following steps:1) Polluted area in video is detected, and is demarcated;2) following mend is carried out to each two field picture in video and paints operation:21) take and current frame image continuously preceding L two field pictures and rear R two field pictures;22) according to current frame image and preceding L two field pictures, the time continuity of rear R two field pictures, using the method for minimizing total variance, the Polluted area in current frame image mend and is painted;23) judge whether the Polluted area of current frame image is mended to paint completely, if it is, entering step 25);If it is not, then it enters step 24);24) utilize the spatial continuity of current frame image itself, using the method for minimizing total variance, in current frame image Polluted area it is remaining do not mend the region painted and mend paint:25) benefit for terminating current frame image is painted.The content benefit of Polluted area in the video of the present invention paints method, and it is higher that benefit paints accuracy.
Description
【Technical field】
The present invention relates to computer video process fields, are painted more particularly to a kind of content benefit of the Polluted area in video
Method.
【Background technology】
There are the region where some Polluted areas, such as station symbol in video, some objects are strayed into picture when shooting video
Region where afterwards, the presence of these Polluted areas both reduced the ornamental comfort level of video, also increases and exchanges between video
Difficulty.Therefore, how to remove Polluted area and accurate mend that paint Polluted area be a key issue urgently to be resolved hurrily.At present
It is usually simply taken when having and Polluted area is carried out by time continuity and spatial continuity to mend the method painted, but repairing
The average of respective pixel point is repaired in front and rear frame reference picture.The repair process is simple and efficient, but accuracy is relatively low.
【The content of the invention】
The technical problems to be solved by the invention are:Above-mentioned the deficiencies in the prior art are made up, propose the dirt in a kind of video
The content benefit in dye region paints method, and it is higher that benefit paints accuracy.
The technical issues of of the invention, is solved by following technical solution:
A kind of content benefit of Polluted area in video paints method, comprises the following steps:1) contaminated area in video is detected
Domain, and demarcated;2) following mend is carried out to each two field picture in video and paints operation:21) take and the continuous preceding L of current frame image
Two field picture and rear R two field pictures;Wherein, L and R is the integer more than or equal to 0, and is 0 during L with R differences, and specific value is by user
Required precision is painted according to benefit to be set;22) according to current frame image and preceding L two field pictures, the time continuity of rear R two field pictures,
Using the method for minimizing total variance, the Polluted area in current frame image mend and is painted;23) dirt of current frame image is judged
Whether dye region, which mends, is painted completely, if it is, entering step 25);If it is not, then it enters step 24);24) present frame figure is utilized
As the spatial continuity of itself, using the method for minimizing total variance, do not mend what is painted to remaining in current frame image Polluted area
Region mend and paints:25) benefit for terminating current frame image is painted.
The advantageous effect that the present invention is compared with the prior art is:
The content benefit of Polluted area in the video of the present invention paints method, from time continuity, is always become using minimizing
The method of difference to the Polluted area in current frame image mend and painted.After being painted using time continuity benefit, such as exist not mend and paint
Region, continue by spatial coherence, carry out supplement and mend to paint to not mending the region painted using the method for total variance is minimized.Due to
Front and rear L, R two field picture is continuous, i.e., front and rear two frames variation is linear, then in addition to Polluted area, total change of front and rear two frame
Poor (total variation) is very small, and in Polluted area, total variance is larger.Similarly, each picture inside current frame image
Vegetarian refreshments is also continuous, and only discontinuous in Polluted area, total variance is larger.Therefore, mend during painting, by minimizing total variance
Method, as successive ignition benefit is painted so that frame and frame, row and row and total variance between the column and the column reach minimum, then it represents that change
It is more continuous for rear current frame image and front and rear two field picture, it is relatively more continuous between each pixel inside current frame image, then reach
The purpose of removal noise, Polluted area benefit is painted preferably.In this way, the method by minimizing total variance, continuous iteration can obtain
It is painted to benefit than more complete two field picture, so as to effectively improve the accuracy for mending process of painting.
【Description of the drawings】
Fig. 1 is that the content benefit of the Polluted area in the video of the specific embodiment of the invention paints the flow chart of method.
【Specific embodiment】
With reference to embodiment and compare attached drawing the present invention is described in further details.
As shown in Figure 1, paint the flow chart of method for the content benefit of the Polluted area in video in present embodiment.It mends
The video image of the input of method processing is painted, Polluted area is as it can be seen that the data format of video is unlimited.Benefit paints method including following
Step:
1):Detect the Polluted area in video, and marking contaminated region.
It during detection, can be detected by artificially observing, alternatively, inspection is identified according to the color change of adjacent two frame
It surveys.During calibration, by the pixel value of the pixel in Polluted area labeled as 1, the pixel value mark of the pixel in uncontaminated region
It is denoted as 0.In this way, establish matrix of the present frame on Polluted area position information, subsequently carry out mending when painting, need to will only repair
Information in result afterwards corresponding to Polluted area is assigned to the Polluted area in original image.
2):Following mend is carried out to each two field picture in video and paints operation:
21) take and current frame image continuously preceding L two field pictures and rear R two field pictures;Wherein, L and R is more than or equal to 0
Integer, and be 0 during L with R differences, specific value is painted required precision according to benefit by user and is set.
Preferably, when taking preceding L two field pictures, rear R two field pictures, when running into cutaway frame or the value of L, R increase to threshold
Stop during value.In this way, not only can ensure that the quantity of the reference frame image of extraction was more, but also it is unlikely to cover the influence of cutaway frame
Follow-up benefit paints precision.
22) according to current frame image and preceding L two field pictures, the time continuity of rear R two field pictures, using minimum total variance
Method to the Polluted area in current frame image mend and painted.
In the step, specifically include:22a), according to the preceding L frames, rear R two field pictures and current frame image, structural matrix
K;Wherein, matrix K includes m rows, n column datas, and the value of m is the number of the pixel included in a two field picture, n L+R+1, matrix
The value of each element of middle same row corresponds to the pixel value of each pixel in a two field picture respectively;
22b) according to the method for minimizing total variance, matrix K is decomposed into matrix T and sparse matrix N, one in matrix T
Column element corresponds to the current frame image mended after painting, and the element for having numerical value in matrix N corresponds to the pixel of the pixel in Polluted area
Value;
22c) use pixel value of the value of respective element in matrix T as the pixel in the Polluted area of current frame image
Mend and paint.
In above-mentioned steps, preceding L two field pictures, the convex image of present frame and the rear R two field pictures overall situation are turned into a matrix, wherein
A column element in each two field picture homography, if matrix is Km*n, m is the size of a two field picture namely the pixel included
Number, n=L+R+1, for the frame number of all two field pictures, the value of each element is the pixel value of respective pixel point, 0~255
In the range of.Since front and rear L+R frames are continuous, i.e., the variation between front and rear two field picture is linear, then except Polluted area
Outside, the total variance (total variation) of front and rear two frame is very small, and in Polluted area, total variance is larger, therefore mends and paint
When, it can will mend and paint process of the process model building as a minimum total variance.Matrix K is decomposed into matrix T and sparse matrix N,
Solving has the element of numerical value containing the pixel information for mending the current frame image after painting in obtained matrix T in obtained matrix N
The pixel information of Polluted area in corresponding current frame image, the value of element is also in the range of 0~255.It therefore, subsequently can be straight
Connect the pixel information that Polluted area in current frame image is painted using the information benefit in matrix T.
The process of above-mentioned minimum total variance iteration can be converted into iterative solution equation below:
min||T||tvSt.K=T+ | | N | |1,
Stopping criterion for iteration is K-T- ‖ N ‖1Reach upper limit value less than threshold value or iterations;Wherein, tv representing matrixes T
Along the total variance of line direction.‖N‖1The first normal form of representing matrix N.The method for solving as above equation is more, such as augmentation glug
Bright day method, it is numerous to list herein.
For example, the pixel number included in each two field picture is 64, L=3, R=3 two field pictures, then such as preceding institute are extracted
Stating the matrix K of construction includes 64 rows, 7 column elements, and each column element corresponds to the pixel value of each pixel in a two field picture.Assuming that structure
It is arranged in order in sequence when making, the first column element corresponds to the preceding first two field picture L1 of extraction, and the second column element corresponds to
The preceding second two field picture L2 of extraction, the 3rd column element correspond to the preceding 3rd two field picture L3 of extraction, and the 4th column element corresponds to present frame
Image Z, the five, the six and seven column elements correspond to the rear first two field picture R1, rear second two field picture R2 and the rear 3rd of extraction respectively
Two field picture R3.After minimizing total variance iterative solution, obtained matrix T and sparse matrix N also include 64 rows, 7 column elements.
Matrix T is along the total variance namely two field picture L1 of line direction and the variation of two field picture L2, the variation of two field picture L2 and two field picture L3,
The variation of two field picture L3 and two field picture Z, the variation of two field picture Z and two field picture R1, the variation of two field picture R1 and two field picture R2, frame figure
As the variation of R2 and two field picture R3, the summation being respectively deteriorated.The variation of a direction is the pixel value of adjacent two pixel in this direction
Subtract each other.It is so that the minimum total variation to mend the target painted.Due to mending the method that the method painted is a global optimization, total variance
Minimum is to act on the L+R+1 two field pictures of extraction, for ensureing the continuity of each two field picture.When benefit paint iteration it is abundant after, obtain
The matrix T-phase ratio arrived reduces the lofty property between the pixel of front and rear two field picture, the pixel of each two field picture in original image frame
It is more continuous between point, so as to achieve the purpose that remove noise.Also after therefore the i.e. corresponding benefit of the 4th column element is painted in matrix T
Current frame image, the situation after remaining six column element corresponds to each two field picture L1, L2, L3 and R1 respectively, R2, R3 benefit are painted.It is sparse
The value of Partial Elements (element for having numerical value) corresponds to the pixel of Polluted area in current frame image in the 4th column element in matrix N
Pixel value, the value of remaining element and remaining six column element is 0 in the 4th column element.After iterative solution obtains matrix T, ginseng
Examine abovementioned steps 1) in mark Polluted area, use in the 4th column element in matrix T correspond to Polluted area position element
Value substitute the pixel value of the pixel in current frame image Z in Polluted area.For example, in foregoing calibration matrix 1
Matrix T corresponding elements are covered respective pixel point in original image frame by the position at place;If 0, then directly skip, so as to carry out
The benefit of current frame image Polluted area is painted.
23) judge whether the Polluted area of current frame image is mended to paint completely, if it is, entering step 25).If not,
It then enters step 24).
24) spatial continuity of current frame image itself is utilized, using the method for minimizing total variance, to current frame image
In Polluted area it is remaining do not mend the region painted and mend paint.
25) benefit for terminating current frame image is painted.
It has such as mended and has been painted completely mentioned by time continuity, then the benefit that can directly terminate current frame image is painted, under
The benefit of one frame is painted.As do not mend also paint it is complete, then by the continuity of the spatial information of present frame itself, using minimizing total variance
Method to residual contamination region carry out mend paint.
It specifically includes:24a) according to current frame image, structural matrix M;Wherein, matrix M includes r rows, c column datas, the value of r
For the columns that the line number that pixel in current frame image is distributed, the value of c are distributed for pixel in current frame image, the value of each element
The pixel value of each pixel in current frame image is corresponded to respectively;
24b) according to the method for minimizing total variance, matrix M is decomposed into matrix P and sparse matrix Q, matrix P, which is corresponded to, to be mended
Current frame image after painting has in sparse matrix Q the element of numerical value to correspond to residue and does not mend pixel in the Polluted area painted
Pixel value;
It is remaining in 24c) using the value of respective element in matrix P as the Polluted area of current frame image not mend the region painted
The pixel value of interior pixel mend and painted.
During above-mentioned minimum total variance, it will wait to mend the integrally constructed matrix M of current frame image itself paintedr *c, r is the row of matrix, and c is matrix column.Since image is continuous on uncontamination regional space, total variance is smaller, and
Polluted area is discontinuous, and total variance is larger, therefore similary further to Polluted area progress using the method for minimizing total variance
Benefit is painted.When benefit is painted, it can will mend and paint process of the process model building as a minimum total variance.Matrix M is decomposed into matrix P and dilute
It dredges to contain in matrix Q, the matrix P solved and be believed by the further pixel for mending the current frame image after painting of spatial continuity
It ceases, the element for having numerical value in obtained matrix Q corresponds to the remaining pixel information for not mending the Polluted area painted in current frame image.
Therefore, the benefit of the information in matrix P subsequently can be used directly and paint in current frame image the remaining region painted of not mending in Polluted area
Pixel information.
Similarly, the process of above-mentioned minimum total variance iteration can be converted into iterative solution equation below:
min||P||tvSt.M=P+ | | Q | |1,
Stopping criterion for iteration is M-P- ‖ Q ‖1Reach upper limit value less than threshold value or iterations;Wherein, tv representing matrixes P
Along the total variance of line direction.‖Q‖1The first normal form of representing matrix Q.The method for solving as above equation is more, such as augmentation glug
Bright day method, it is numerous to list herein.
Similarly, for example, the pixel number included in each two field picture is a for 64 (8*8), then the square constructed as previously described
Battle array M includes 8 rows, 8 column elements, and each element corresponds to the pixel value of each pixel in current frame image respectively.Polluted area corresponds to left
3*3=9 pixel I1, I2, I3, I4, I5, I6, I7, I8 and the I9 at upper angle.It is foregoing to have painted 5 pixels through time continuity benefit
Point I3, I6, I7, I8 and I9, the remaining upper left corner 2*2=4 pixel I1, I2, I4, I5, which are not mended, to be painted.
After minimizing total variance iterative solution, obtained matrix P and sparse matrix Q also include 8 rows, 8 column elements.Square
For battle array P along the variation between the total variance of line direction namely the first column element and the second column element, the second column element and the 3rd row are first
Variation between element, the variation ... ... between the 3rd column element and the 4th column element, between the 7th column element and the 8th column element
Variation, the summation being respectively deteriorated.It is so that the minimum total variation to mend the target painted.It is that an overall situation is excellent due to mending the method painted
The method of change, minimum total variation are to act on entire image, for ensureing the piecewise smooth of image.When benefit, to paint iteration abundant
Afterwards, the matrix P obtained reduces the lofty property between pixel compared to original image frame, and pixel color is more continuous, so as to
Achieve the purpose that remove residual noise.Also therefore 4, upper left corner element is corresponded in the current frame image mended after painting in matrix P
The pixel value of 4 elements in the upper left corner, remaining 60 element are close with the pixel value of 60 elements of remaining in current frame image respectively
Seemingly.The value of 4 elements in the upper left corner corresponds to the remaining pixel for not mending the Polluted area painted in current frame image in sparse matrix Q
Pixel value, the value of remaining 60 element is close to 0.After iterative solution obtains matrix P, do not mend what is painted with reference to pre-determined residue
Polluted area is 4, upper left corner element, then using residue in the value replacement current frame image Z of 4, the upper left corner in matrix P element not
The pixel value of the pixel in the Polluted area painted is mended, completion current frame image residue, which is not mended, paints the further of Polluted area
Benefit is painted.
It is complete using the method for minimizing total variance by the information of above-mentioned time continuity and the information of spatial continuity
Content benefit into the Polluted area of current frame image is painted.Carry out successively each two field picture benefit paint to get to it is all benefit paint completion
Each two field picture.Each two field picture that all benefits paint completion is integrated sequentially in time, you can output obtains removal contaminated area
The video that domain obtains after mending and painting.
The content benefit of Polluted area in the video of present embodiment paints method, by the side for minimizing total variance
Method, as successive ignition benefit is painted so that total variance of the frame with frame, row and row and between the column and the column reaches minimum, then it represents that during iteration
Polluted area, which has been mended, to be painted preferably, and front and rear more continuous.In this way, the method by minimizing total variance, continuous iteration can obtain
It is painted to benefit than more complete two field picture, so as to effectively improve the accuracy for mending process of painting.
The above content is a further detailed description of the present invention in conjunction with specific preferred embodiments, it is impossible to assert
The specific implementation of the present invention is confined to these explanations.For those of ordinary skill in the art to which the present invention belongs, exist
Several replacements or apparent modification are made on the premise of not departing from present inventive concept, and performance or purposes are identical, should all be considered as
It belongs to the scope of protection of the present invention.
Claims (8)
1. a kind of content benefit of the Polluted area in video paints method, it is characterised in that:Comprise the following steps:
1) Polluted area in video is detected, and is demarcated;
2) following mend is carried out to each two field picture in video and paints operation:
21) take and current frame image continuously preceding L two field pictures and rear R two field pictures;Wherein, L and R is the integer more than or equal to 0,
And for 0 during L with R differences, specific value is painted required precision according to benefit by user and is set;
22) according to current frame image and preceding L two field pictures, the time continuity of rear R two field pictures, using the side for minimizing total variance
Method to the Polluted area in current frame image mend and painted;It specifically includes:22a) according to the preceding L frames, rear R two field pictures and
Current frame image, structural matrix K;Wherein, matrix K includes m rows, n column datas, and the value of m is the pixel included in a two field picture
Number, n L+R+1, the value of each element of same row corresponds to the pixel value of each pixel in a two field picture respectively in matrix;
22b) according to the method for minimizing total variance, matrix K is decomposed into matrix T and sparse matrix N, the column element pair in matrix T
The current frame image after painting should be mended, the value for having the element of numerical value in matrix N corresponds to the pixel value of the pixel in Polluted area;Its
In, the method for minimizing total variance is iterative solution equation below:min||T||tvSt.K=T+ | | N | |1, stopping criterion for iteration
For K-T- ‖ N ‖1Reach upper limit value less than threshold value or iterations;Wherein, total variances of the tv representing matrixes T along line direction;
22c) mended using the pixel value of value as the pixel in the Polluted area of current frame image of respective element in matrix T
It paints;
23) judge whether the Polluted area of current frame image is mended to paint completely, if it is, entering step 25);If it is not, then into
Enter step 24);
24) spatial continuity of current frame image itself is utilized, using the method for minimizing total variance, current frame image is polluted
In region it is remaining do not mend the region painted and mend paint:
25) benefit for terminating current frame image is painted.
2. the content benefit of the Polluted area in video according to claim 1 paints method, it is characterised in that:It is drawn by augmentation
Ge Lang methods solve equation.
3. the content benefit of the Polluted area in video according to claim 1 paints method, it is characterised in that:The step
24) include:24a) according to current frame image, structural matrix M;Wherein, matrix M includes r rows, c column datas, and the value of r is present frame
The line number that pixel is distributed in image, the value of c are the columns that pixel is distributed in current frame image, and the value of each element corresponds to respectively
The pixel value of each pixel in current frame image;24b) according to minimize total variance method, by matrix M be decomposed into matrix P and
Sparse matrix Q, matrix P correspond to the current frame image mended after painting, and the value for having the element of numerical value in matrix Q is corresponded in Polluted area
The pixel value of pixel;The value of respective element in matrix P 24c) is used not mended as remaining in the Polluted area of current frame image
The pixel value for the pixel in region painted mend and painted.
4. the content benefit of the Polluted area in video according to claim 3 paints method, it is characterised in that:The step
In 24b), the method for minimizing total variance is iterative solution equation below:min||P||tvSt.M=P+ | | Q | |1, iteration ends
Condition is M-P- | | Q | |1Reach upper limit value less than threshold value or iterations;Wherein, total changes of the tv representing matrixes P along line direction
Difference.
5. the content benefit of the Polluted area in video according to claim 4 paints method, it is characterised in that:It is drawn by augmentation
Ge Lang methods solve equation.
6. the content benefit of the Polluted area in video according to claim 1 paints method, it is characterised in that:The step
21) in:When taking preceding L two field pictures, rear R two field pictures, until stopping when running into cutaway frame or when the value of L, R increase to threshold value.
7. the content benefit of the Polluted area in video according to claim 1 paints method, it is characterised in that:The step 1)
In, during detection, it is detected by artificially observing, alternatively, detection is identified according to the color change of adjacent two frame.
8. the content benefit of the Polluted area in video according to claim 1 paints method, it is characterised in that:Further include step
3):Each two field picture that all benefits paint completion is integrated sequentially in time, output removal Polluted area obtains after mending and painting
Video.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510760914.2A CN105376462B (en) | 2015-11-10 | 2015-11-10 | A kind of content benefit of Polluted area in video paints method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510760914.2A CN105376462B (en) | 2015-11-10 | 2015-11-10 | A kind of content benefit of Polluted area in video paints method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105376462A CN105376462A (en) | 2016-03-02 |
CN105376462B true CN105376462B (en) | 2018-05-25 |
Family
ID=55378231
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510760914.2A Active CN105376462B (en) | 2015-11-10 | 2015-11-10 | A kind of content benefit of Polluted area in video paints method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105376462B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112118483A (en) * | 2020-06-19 | 2020-12-22 | 中兴通讯股份有限公司 | Video processing method, device, equipment and storage medium |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103618905A (en) * | 2013-12-09 | 2014-03-05 | 清华大学深圳研究生院 | Content drawing method for station caption area in video |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8805106B2 (en) * | 2008-09-26 | 2014-08-12 | Futurewei Technologies, Inc. | System and method for compressing and decompressing images and video |
-
2015
- 2015-11-10 CN CN201510760914.2A patent/CN105376462B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103618905A (en) * | 2013-12-09 | 2014-03-05 | 清华大学深圳研究生院 | Content drawing method for station caption area in video |
Non-Patent Citations (2)
Title |
---|
Object Enhancement and Recognition Based on Hough Forest for Underground Video;Bo Fu等;《2014 Sixth International Symposium on Parallel Architectures, Algorithms and Programming (PAAP)》;20140715;全文 * |
基于分块思想的最小化总变差去噪新算法;吴凯凯等;《计算机工程与设计》;20111116;第32卷(第11期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN105376462A (en) | 2016-03-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105261037B (en) | A kind of moving target detecting method of adaptive complex scene | |
CN103268481B (en) | A kind of Text Extraction in complex background image | |
CN107014819A (en) | A kind of solar panel surface defects detection system and method | |
CN104021527B (en) | Rain and snow removal method in image | |
CN103310422B (en) | Obtain the method and device of image | |
CN102968765B (en) | Method for correcting infrared focal plane heterogeneity based on sigma filter | |
Chu et al. | Object tracking algorithm based on camshift algorithm combinating with difference in frame | |
CN103559693A (en) | Image local structure self-adaption recovery method based on non-continuity indicator | |
CN103065334A (en) | Color cast detection and correction method and device based on HSV (Hue, Saturation, Value) color space | |
CN102567722A (en) | Early-stage smoke detection method based on codebook model and multiple features | |
CN104915938A (en) | Restoration method for image with high density impulse noise pollution | |
CN103020921A (en) | Single image defogging method based on local statistical information | |
CN113012096B (en) | Display screen sub-pixel positioning and brightness extraction method, device and storage medium | |
CN102609909A (en) | Method and device for defogging single image | |
CN107749986A (en) | Instructional video generation method, device, storage medium and computer equipment | |
ATE498874T1 (en) | METHOD FOR DETECTING A MOVING OBJECT IN A SEQUENCE OF IMAGES CAPTURED BY A MOVING CAMERA, COMPUTER SYSTEM AND COMPUTER PROGRAM PRODUCT | |
CN105376462B (en) | A kind of content benefit of Polluted area in video paints method | |
Xia et al. | Single image rain removal via a simplified residual dense network | |
CN106327450A (en) | Method for enhancing low-light video image based on space-time accumulation and image degradation model | |
CN105389781A (en) | Vehicle light repairing method based on saliency detection and Criminisi algorithm | |
CN106600539B (en) | A kind of single image super-resolution method based on part across scale autoregression model | |
CN107122715A (en) | It is a kind of based on frequency when conspicuousness combine moving target detecting method | |
JP2005345290A (en) | Streak-like flaw detecting method and streak-like flaw detector | |
Li et al. | Fast visual tracking using motion saliency in video | |
CN112215766B (en) | Image defogging method combining image restoration and image enhancement and convolution network thereof |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |