CN1668068A - An adaptive picture noise suppression method - Google Patents
An adaptive picture noise suppression method Download PDFInfo
- Publication number
- CN1668068A CN1668068A CN 200510049598 CN200510049598A CN1668068A CN 1668068 A CN1668068 A CN 1668068A CN 200510049598 CN200510049598 CN 200510049598 CN 200510049598 A CN200510049598 A CN 200510049598A CN 1668068 A CN1668068 A CN 1668068A
- Authority
- CN
- China
- Prior art keywords
- edge
- pixel
- noise
- filtering
- image
- 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.)
- Granted
Links
Images
Landscapes
- Picture Signal Circuits (AREA)
Abstract
This invention relates to a method for applying an adaptive method to suppress image noises including utilizing the inter-field recursive method to find out edges of the image and present diameter and direction of the edges by computing the brightness correlation of being selected pixel points, utilizing the image blanking period information to estimate the noise grade of the local field image, selecting related filter factors to filter to get the noise reduction result. This invention can apply filter factors with different strength to different pollutant degrees of images to get a more natural image result.
Description
Technical field
The present invention is applied to the Digital Video Processing technical field, particularly a kind of method that adopts adaptive approach to suppress picture noise.
Background technology
In each link of TV signal source generation, Channel Transmission and receiving course, all may introduce noise, the most general TV signal noise type is an additive white Gaussian noise, mainly produces the Channel Transmission process of videoblankingsync from the transmission end to user's receiving terminal.The method that reduces picture noise has a variety of, but most widely used general, realizes the simplyst, and the most tangible method of effect is carried out low-pass filtering to image exactly.Low-pass filtering is mainly utilized image correlation spatially, and random noise and the Gaussian noise relatively poor to those correlations suppress.
In general, image noise reduction can be divided in the frame and two kinds of methods of interframe noise reduction.Noise reduction mainly is to utilize image correlation spatially to handle in the frame, and the interframe noise reduction is to utilize image correlation in time to handle.Their unavoidable weakness is all arranged for two kinds of methods.
For the space noise reduction, i.e. noise reduction in the frame, its effect aspect the handled object border is very not satisfactory.The border brightness and the colourity of object all can have bigger variation, and when noise reduction point just was in object boundary, low pass filter can thicken object boundary, and this is that great majority are realized the problem that noise reduction all can run into by low pass filter.
The method that a kind of noise magnitude that application number is 200410069837.8 disclosure of the Invention is measured.Adopted definitely the adding up and, and select one group of predetermined filter coefficient to carry out filtering of luminance difference of current field picture and preceding field picture respective pixel in this method by this magnitude as the noise magnitude of this field picture.This kind method has effect preferably for rest image, but for moving image, particularly the image of rapid movement is understood the introducing bigger noise magnitude wrong owing to the motion of image itself.The noise suppressed coefficient that adopts this noise magnitude to determine will bring bigger image definition loss.
Summary of the invention
The present invention is directed to the space noise reduction and can produce obscurity boundary, and problem such as noise magnitude measurement, provided a kind of method that image side information and picture blanking phase carry out the Adaptive Suppression noise of video image of utilizing.
The present invention includes following steps: (a) utilize the method for recurrence between the field, by calculating, find the edge of image, and provide the size and Orientation at edge to candidate pixel point brightness correlation; (b) utilize picture blanking phase information that the noisiness level of this field picture is estimated, the noise pollution that is subjected to of university degree field picture is strong more more for the noise magnitude; (c) according to the size and Orientation at edge, select the filtering pixel,, choose corresponding filter factor according to the noise magnitude; (d) utilize the filtering pixel and the corresponding filter factor that produce to carry out filtering, obtain the noise reduction result.
Between the field described in the step (a) recursion method be meant utilize edge result that the front court calculates as the Ben Chang edge judge a foundation, if there is the edge in the front court pixel, then this correspondence position need carry out candidate pixel point brightness correlation calculations and increase the weight that there is the edge in this pixel.
The candidate pixel point brightness correlation calculations in step described in the step (a) is with corresponding candidate pixel luminance difference summation; The result who compares all directions correlation calculations, the direction that correlation is bigger are the directions that the edge exists, and the foundation of calculating as next edge is preserved at this edge.
It is exactly the mean value of computed image blanking interval level that noise magnitude described in the step (b) is estimated, this mean value is as the noise magnitude.
The system of selection of filtering pixel is in the step (c): for the pixel of edge existence, direction according to the edge, select the filtering pixel according to the edge direction, the distance of filtering pixel distance impact point is by the size decision at edge, and edge big more candidate's filtering pixel is far away more apart from impact point; For the non-existent pixel in edge, select the point adjacent as the filtering pixel with object pixel.
The choosing method of filter factor is in the step (c): for the pixel that the edge exists, filter factor is exactly to finish average to the filtering pixel; For the pixel that does not have the edge, choose filter factor according to the size of noise magnitude.
The adaptivity that the invention provides noise-reduction method is mainly reflected in two aspects: (1) is judged by the edge, the candidate pixel of adaptively selected filter.Judgement be the pixel at image edge, and is big with the correlation of the pixel of edge direction, utilize pixel on the edge to carry out filtering and can obtain better filter effect, and minimizing is to the loss of edge details.Judgement is the pixel at non-edge, utilizes the candidate pixel that pre-determines to carry out filtering; (2) judge adaptively selected noise reduction coefficient by the noise magnitude.The noise magnitude is being represented image by the degree of noise pollution, and the image of different pollution levels is adopted the filter factor of varying strength, can obtain more natural image effect.
Description of drawings
Fig. 1 is by the block diagram of noise suppression circuit of the invention process;
The block diagram that the noise magnitude that Fig. 2 the present invention adopts is judged;
Fig. 3 is the realization block diagram that 202 average magnitudes are calculated among Fig. 2;
Fig. 4 is the block diagram that 103 edges that adopt detect among Fig. 1;
Fig. 5 is the schematic diagram of SAD calculated direction shown in 404 among Fig. 4;
Fig. 6 is the structured flowchart that pixel is chosen shown in 104 among Fig. 1.
Embodiment
Fig. 1 provides the structured flowchart of the noise-reduction method that present embodiment adopted, and has mainly comprised 5 parts: the noise magnitude estimates that 101 are used for determining the noise magnitude.Because the channel of video signal transmission and media is different, the performance of receiver also is not quite similar, the difference of surrounding environment in addition, and these factors all can make the magnitude of noise different.Therefore the noise magnitude estimator that adopting present embodiment is provided can be determined the magnitude of noise exactly, according to this noise magnitude, can reasonably determine filter factor, obtains optimum filter effect.102 results according to noise magnitude and edge judgement provide one group of optimized coefficient of determining through experiment.Edge calculator shown in 103 is used for the edge of computed image, and its computational process is specifically set forth by Fig. 4.103 side informations that obtain that calculate offer 104 pixels and choose module, and 104 algorithm structures that pass through self judge that 103 calculate the accuracy at edges.Adopt different pixels to choose scheme for the pixel that is judged as edge and non-edge, selected pixel is delivered to afterbody 105 filters.104 also to give the filter factor that 102,102 results that judge in view of the above provide edge noise reduction or non-edge noise reduction accordingly simultaneously the result that the edge is judged.102 and 104 candidate pixels that obtain and candidate coefficient will be as 105 inputs.105 with candidate pixel and coefficient of correspondence multiplies each other and to product summation, obtain filter output, and this output is noise reduction output.
Shown in Figure 2 is 101 refined structure figure among Fig. 1.201 is window choice devices among the figure.The effect of this device is to produce the valid window that a noise magnitude is calculated.What present embodiment provided is a kind of a kind of method that the noise magnitude is estimated of carrying out based on the blanking interval level fluctuation.When image was by noise pollution, blanking interval can be subjected to the pollution of noise equally, and was identical on magnitude.Level at the ideal situation blanking interval should remain unchanged, and the variation of blanking interval level has obviously embodied The noise, thus the blanking period calculating of carrying out the noise magnitude can to remove the estimation that is caused by image motion inaccurate.201 are just providing the valid window in blanking interval calculating noise magnitude, under the restriction of this window, extracting the average blanking interval level that valid data and 202 obtain from the video data of input subtracts each other through 203,204 take absolute value, 205 add up three processes obtain absolute difference accumulative total and, this numerical value is exactly the noise magnitude when the front court.
Fig. 3 is to 202 detailed description among Fig. 2.202 is average blanking level computing modules among Fig. 2, and its effect is the reference level that calculates blanking interval, and this reference level is a benchmark of calculating noise magnitude, and it is the calculated value that adds up on average to obtain through to multiframe blanking interval level.Reference level is the numerical value of kept stable, and he can often not change with incoming frame, also Just because of this just can be with it as reference level.The acquisition of the reference level of present embodiment obtains by multiframe signal is averaged, and it can all not upgrade by each frame, only behind the frame of accumulative total some, just can upgrade.The reference level that the accumulation of process multiframe calculates is only accurately.301 is synchronization decisions devices that detect synchronizing signal, and it is judged to the frame/field mark that makes new advances by the polarity and the width characteristics of synchronizing signal.Count these 302 pairs of incoming frame/fields of sign indication, and when frame/field of input ran up to a certain value TH that presets, update signal of 303 renewal decision devices generations was upgraded reference level.304 and 305 is the circuit that calculate reference level in the present embodiment, and it averages the value through the blanking level of 301 gatings and obtains average a reference level.
Fig. 4 is the block diagram of the edge computing module of present embodiment employing.Present embodiment has adopted a kind of edge computational methods of recurrence.401 is edge memories, and it has preserved the side information of Shang Yichang/frame.The 404th, the major part that the edge calculates, it passes through the calculating to the SAD of the pixel of all directions, obtains one group of information that most possibly reflects image edge direction.The weighted value of all directions that the SAD of 404 all directions that obtain and 402 obtains is weighted comparison at 405 kinds, obtains the edge direction of maximum likelihood.
Fig. 5 has provided the schematic diagram of edge calculated direction and SAD calculating.Among the figure coordinate be (i, pixel j) are object pixel, brightness value be designated as f (i, j), the pixel of adjacent lines above the j-1 representative, setting m is a candidate direction, if get m=[-6 ,+6], the sad value of direction m can be provided by following expression:
As the SAD of m=0 interval scale vertical direction, representing the left and right sides scope when m gets other values respectively is 6 direction.Can obtain the SAD of 13 candidate direction thus.SAD
mWorth size has been represented the size that has the possibility at edge on direction m, SAD
mBe worth more for a short time, be illustrated on the direction of m and exist the possibility at edge big more.
104 at first judge 103 edges that calculate among Fig. 1.Be judged as really and will select candidate pixel to carry out the filtering of edge direction according to the edge direction, but not then carry out common space filtering for the pixel at edge for the impact point at edge.Fig. 6 has introduced a kind of realization block diagram of this process.
Claims (6)
1, a kind of method of adaptive picture noise suppression is characterized in that this method may further comprise the steps: (a) utilize the method for recurrence between the field, by the calculating to candidate pixel point brightness correlation, find the edge of image, and provide the size and Orientation at edge; (b) utilize picture blanking phase information that the noisiness level of this field picture is estimated, the noise pollution that is subjected to of university degree field picture is strong more more for the noise magnitude; (c) according to the size and Orientation at edge, select the filtering pixel,, choose corresponding filter factor according to the noise magnitude; (d) utilize the filtering pixel and the corresponding filter factor that produce to carry out filtering, obtain the noise reduction result.
2, the method for a kind of adaptive picture noise suppression as claimed in claim 1, it is characterized in that recursion method between the field described in the step (a) be meant utilize edge result that the front court calculates as the Ben Chang edge judge a foundation, if there is the edge in the front court pixel, then this correspondence position need carry out candidate pixel point brightness correlation calculations and increase the weight that there is the edge in this pixel.
3, the method for a kind of adaptive picture noise suppression as claimed in claim 1, the candidate pixel point brightness correlation calculations that it is characterized in that the step described in the step (a) are with corresponding candidate pixel luminance difference summation; The result who compares all directions correlation calculations, the direction that correlation is bigger are the directions that the edge exists, and the foundation of calculating as next edge is preserved at this edge.
4, the method for a kind of adaptive picture noise suppression as claimed in claim 1 is characterized in that it is exactly the mean value of computed image blanking interval level that the noise magnitude described in the step (b) is estimated, this mean value is as the noise magnitude.
5, the method for a kind of adaptive picture noise suppression as claimed in claim 1, the system of selection that it is characterized in that filtering pixel in the step (c) is: for the pixel of edge existence, direction according to the edge, select the filtering pixel according to the edge direction, the distance of filtering pixel distance impact point is by the size decision at edge, and edge big more candidate's filtering pixel is far away more apart from impact point; For the non-existent pixel in edge, select the point adjacent as the filtering pixel with object pixel.
6, the method for a kind of adaptive picture noise suppression as claimed in claim 1 is characterized in that the choosing method of filter factor is in the step (c): for the pixel that the edge exists, filter factor is exactly to finish average to the filtering pixel; For the pixel that does not have the edge, choose filter factor according to the size of noise magnitude.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100495984A CN100367771C (en) | 2005-04-08 | 2005-04-08 | An adaptive picture noise suppression method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100495984A CN100367771C (en) | 2005-04-08 | 2005-04-08 | An adaptive picture noise suppression method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1668068A true CN1668068A (en) | 2005-09-14 |
| CN100367771C CN100367771C (en) | 2008-02-06 |
Family
ID=35038933
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005100495984A Active CN100367771C (en) | 2005-04-08 | 2005-04-08 | An adaptive picture noise suppression method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100367771C (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100505832C (en) * | 2006-03-21 | 2009-06-24 | 中国科学院计算技术研究所 | Image de-noising process of multi-template mixed filtering |
| CN101536031A (en) * | 2006-09-29 | 2009-09-16 | 汤姆森许可贸易公司 | Automatic parameter estimation for adaptive pixel-based filtering |
| CN101938597A (en) * | 2009-06-30 | 2011-01-05 | 株式会社日立制作所 | Image recording system |
| CN101018289B (en) * | 2006-02-07 | 2012-05-30 | 冲电气工业株式会社 | Device for measuring amount of noise |
| CN101261734B (en) * | 2007-03-06 | 2012-09-05 | 佳能株式会社 | Image treatment device and image treatment method |
| CN103414845A (en) * | 2013-07-24 | 2013-11-27 | 中国航天科工集团第三研究院第八三五七研究所 | Self-adaptive video image noise reducing method and noise reducing system |
| CN104143177A (en) * | 2013-05-07 | 2014-11-12 | 江南大学 | Method for eliminating image interference of line-scan digital camera |
| CN104168405A (en) * | 2013-05-20 | 2014-11-26 | 聚晶半导体股份有限公司 | Noise reduction method and image processing device |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1419680A (en) * | 2001-01-26 | 2003-05-21 | 皇家菲利浦电子有限公司 | Spatio-temporal filter unit and image display apparatus comprising such a spatio-temporal filter unit |
| KR100522607B1 (en) * | 2003-07-15 | 2005-10-19 | 삼성전자주식회사 | Apparatus and method for executing adaptive video signal processing according to noise condition |
| KR100497398B1 (en) * | 2003-07-16 | 2005-06-23 | 삼성전자주식회사 | Method and apparatus of video noise reduction |
-
2005
- 2005-04-08 CN CNB2005100495984A patent/CN100367771C/en active Active
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101018289B (en) * | 2006-02-07 | 2012-05-30 | 冲电气工业株式会社 | Device for measuring amount of noise |
| CN100505832C (en) * | 2006-03-21 | 2009-06-24 | 中国科学院计算技术研究所 | Image de-noising process of multi-template mixed filtering |
| CN101536031A (en) * | 2006-09-29 | 2009-09-16 | 汤姆森许可贸易公司 | Automatic parameter estimation for adaptive pixel-based filtering |
| CN101261734B (en) * | 2007-03-06 | 2012-09-05 | 佳能株式会社 | Image treatment device and image treatment method |
| CN101938597A (en) * | 2009-06-30 | 2011-01-05 | 株式会社日立制作所 | Image recording system |
| CN101938597B (en) * | 2009-06-30 | 2013-06-12 | 株式会社日立制作所 | Video recording system |
| CN104143177A (en) * | 2013-05-07 | 2014-11-12 | 江南大学 | Method for eliminating image interference of line-scan digital camera |
| CN104168405A (en) * | 2013-05-20 | 2014-11-26 | 聚晶半导体股份有限公司 | Noise reduction method and image processing device |
| CN104168405B (en) * | 2013-05-20 | 2017-09-01 | 聚晶半导体股份有限公司 | Noise suppressing method and its image processing apparatus |
| CN103414845A (en) * | 2013-07-24 | 2013-11-27 | 中国航天科工集团第三研究院第八三五七研究所 | Self-adaptive video image noise reducing method and noise reducing system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100367771C (en) | 2008-02-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1668068A (en) | An adaptive picture noise suppression method | |
| US7570309B2 (en) | Methods for adaptive noise reduction based on global motion estimation | |
| EP1681849B1 (en) | Apparatus for removing noise from a video signal | |
| US7286716B2 (en) | Image quality enhancement circuit and method using inter-frame correlativity | |
| CN101836441B (en) | Method, apparatus, and system for line-based motion compensation in video image data | |
| US20060139494A1 (en) | Method of temporal noise reduction in video sequences | |
| CN104980626A (en) | Method And Apparatus For Reducing Noise Of Image | |
| US20020171759A1 (en) | Adaptive interlace-to-progressive scan conversion algorithm | |
| US20030052979A1 (en) | Image processing method and apparatus | |
| CN102946505B (en) | Self-adaptive motion detection method based on image block statistics | |
| CN106210448B (en) | Video image jitter elimination processing method | |
| CN103179325A (en) | Self-adaptive 3D (Three-Dimensional) noise reduction method for low signal-to-noise ratio video under fixed scene | |
| US7961972B2 (en) | Method and apparatus for short range motion adaptive noise reduction | |
| CN1383673A (en) | Noise filtering image sequence | |
| Zlokolica et al. | Noise estimation for video processing based on spatio-temporal gradients | |
| CN108270945B (en) | Motion compensation denoising method and device | |
| KR20060093074A (en) | Global motion adaptive system with motion values correction with respect to luminance level | |
| CN102447870A (en) | Detection method for static objects and motion compensation device | |
| EP2107523A1 (en) | Method for detecting scene change in a video picture sequence | |
| US8576337B2 (en) | Video image processing apparatus and video image processing method | |
| US8090210B2 (en) | Recursive 3D super precision method for smoothly changing area | |
| CN1199449C (en) | Method for removing intersection by adopting error protection motion compensation and its equipment | |
| US7365799B2 (en) | Gradation display device | |
| Park et al. | Covariance-based adaptive deinterlacing method using edge map | |
| CN102497523B (en) | Median-filtering de-interlacing method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C56 | Change in the name or address of the patentee | ||
| CP03 | Change of name, title or address |
Address after: No. 90, Wensanlu Road, Hangzhou, Zhejiang, Xihu District Patentee after: Hangzhou National Chip Science & Technology Co., Ltd. Address before: 5A, Neusoft building, No. 99 Huaxing Road, Hangzhou, Zhejiang, Xihu District Patentee before: Hangzhou Guoxin Science & Technology Co., Ltd. |