CN1514649A - Image pretreatment method used for video frequency signal compression - Google Patents
Image pretreatment method used for video frequency signal compression Download PDFInfo
- Publication number
- CN1514649A CN1514649A CNA031533450A CN03153345A CN1514649A CN 1514649 A CN1514649 A CN 1514649A CN A031533450 A CNA031533450 A CN A031533450A CN 03153345 A CN03153345 A CN 03153345A CN 1514649 A CN1514649 A CN 1514649A
- Authority
- CN
- China
- Prior art keywords
- image
- filtering
- matching error
- low pass
- estimation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000007906 compression Methods 0.000 title claims description 13
- 230000006835 compression Effects 0.000 title claims description 11
- 238000002203 pretreatment Methods 0.000 title claims description 11
- 238000001914 filtration Methods 0.000 claims abstract description 50
- 238000000034 method Methods 0.000 claims description 36
- 238000007781 pre-processing Methods 0.000 claims description 11
- 230000008569 process Effects 0.000 claims description 10
- 230000006978 adaptation Effects 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 abstract description 2
- 238000002474 experimental method Methods 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 206010019133 Hangover Diseases 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000002560 therapeutic procedure Methods 0.000 description 1
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention includes following steps: (1) combining motion estimation with time-domain filtering; (2) carrying out motion estimation of video signal by using block of current video image as a unit; (3) based on matching error of matched blocks in motion estimation, determining filtering characteristic of low pass filter; (4) carrying out time-domain filtering for 'blocks to be encoded'; (5) time-domain low pass filtering matched blocks in motion estimation of followed-up coded rebuilt images; (6) time-domain filtering 'blocks to be encoded'. Based on motion estimation and time-domain filtering, the invention carries out blocks of image and rebuilds coded image and adjusts filtering intensity in self-adaptation based on matching error. Experiment shows that the invention provides high quality in subjective assessment under certain transmission speed, and reduced compact code rate under same quality of image.
Description
Affiliated technical field
The present invention relates to the compression process field and the vision signal noise filtering field of video communication field, vision signal.More particularly, the present invention can be used for video telephone, video conferencing, video monitoring, a kind of image preprocessing method of video signal compression process field such as video information storage.
Technical background
In the compression of vision signal is handled, manage to reduce the noise of incoming video signal at present, to improve compression efficiency of video signal and to improve subjective picture quality, the method that adopts has time-domain filtering and two kinds of methods of airspace filter usually.Wherein time-domain filtering often adopts the multiple image weighted-average method, and Fig. 1 shows the position of general time domain filtering in encoding video pictures.As can be seen from the figure, this method is not considered the adjacent image difference that video motion causes.Though adopt this time-domain filtering method better to the noise filtering effect of static video image, it may cause hangover to motion video image, cause the subjective quality of image to descend.And airspace filter adopts in piece image the neighbor weighted-average method more, the position of this airspace filter device in encoding video pictures as shown in Figure 2, the smooth image that this airspace filter method is less to details, filtering noise effectively, but edge of image and detail textures are had damage.Another weak point of airspace filter device is that amount of calculation is bigger, and is also higher to the requirement of hardware, DSP disposal ability in actual applications.For now, the method for also pre-treatment not being stipulated in the existing image compression encoding framework.
Summary of the invention
The objective of the invention is provides a kind of computing simple at the deficiencies in the prior art part, be easy on DSP or hardware, realize, effective filtering noise, improve the compression efficiency of image, can reduce streaking distortion and image blurring image preprocessing method according to the kinetic characteristic of vision signal again.
To achieve these goals, the present invention proposes a kind of image preprocessing method that is used for video signal compression: a. estimation and general time-domain filtering is combined; B. be that the motion match piece of unit subregion and coding and rebuilding image carries out the time domain low-pass filtering with the piece to original image; C. adjust the filtering strength in different images zone according to the matching error self adaptation of estimation match block; D. image pre-treatment process is finished in the image encoding process simultaneously.It comprises the following step: 1. be the estimation that unit carries out vision signal with the piece; 2. determine the filtering characteristic of low pass filter according to the matching error of estimation match block; 3. " block to be encoded " carried out time-domain filtering.
The cut-off frequency of described time domain low pass filter is relevant with matching error, and matching error is big, the filter cut-off frequency height, matching error is little, filter cut-off frequency is low, and image pre-treatment process is finished in the image encoding process simultaneously, and the position of this image pre-processing module in video signal coding as shown in Figure 3.
The matching error according to the estimation match block described in the present invention is determined the low pass filter filtering characteristic, mainly comprises: to luminance component is to determine the low pass filter cut-off frequency according to the matching error of brightness match block; Chromatic component is determined the low pass filter cut-off frequency according to the matching error of chromaticity match piece.
Filtering described in the present invention is carried out between original image and previous frame coding and rebuilding image.
Filtering described in the present invention is to carry out between the motion match piece of original image and coding and rebuilding image.
The low pass filter cut-off frequency of the different images piece described in the present invention is relevant with matching error;
The present invention has following useful good effect than prior art:
The present invention is in conjunction with the filtering principle of estimation and general time domain filtering, be deep in the process of general image compressed encoding, do not change the framework of image compression encoding, use some pilot processs of image encoding framework, time-domain filtering is carried out in original image subregion and coding and rebuilding image.One side is filtering noise effectively, improve the compression efficiency of image, again can be according to the kinetic characteristic of vision signal, adjust the filtering strength in different images zone according to the matching error self adaptation, reduce the streaking distortion that general time-domain filtering may cause, nor can produce in the general airspace filter device the level and smooth image blurring effect that causes owing to adjacent image point.This in addition method computing is simple, is easy to realize on DSP or hardware, can not cause too much computational burden to encoder.The present invention is that the difference between original image and the reconstructed image is encoded owing to fully taking into account coding, between original image and coding and rebuilding image, carry out filtering and adopt, therefore can reduce code check very effectively, this method has been carried out the filtering of varying strength to different image-regions in addition, thereby can under the prerequisite that guarantees abundant filtering noise, reduce the distortion of streaking.The operand that the present invention increases in video signal compression is very little, is convenient to realize, under given transmission rate, compressed image subjective assessment quality is significantly improved, under the identical image quality, code check significantly reduces.In actual applications, to static relatively video image, encoder bit rate can be reduced to original 20%-30%, video image to little motion, encoder bit rate can be reduced to original 50%-70%, and for the video image of middle motion, encoder bit rate can hang down original 70%-80%, to big motion video image, encoder bit rate can be reduced to original 80%-90%.
The present invention has following characteristics: a. estimation and general time-domain filtering is combined; B. be that the motion match piece of unit subregion and coding and rebuilding image carries out the time domain low-pass filtering with the piece to original image; C. adjust the filtering strength in different images zone according to the matching error self adaptation of estimation match block, matching error is big, the filter cut-off frequency height, and matching error is little, and filter cut-off frequency is low; D. image pre-treatment process is finished in the image encoding process simultaneously, and the operand of increase is very little.
Description of drawings
Fig. 1 is that the position of known time domain filtering in image encoding concerns block diagram.
Fig. 2 is that the position of known airspace filter device in image encoding concerns block diagram.
Fig. 3 is that the position of image pre-treatment filter of the present invention in image encoding concerns block diagram.
Fig. 4 is that image pre-treatment filter of the present invention is realized theory diagram
Specific embodiment
In order to further specify rather than limit above-mentioned way of realization of the present invention, provide most preferred embodiment below in conjunction with accompanying drawing, thereby make details of the present utility model and advantage become more obvious.
Fig. 1, Fig. 2 are that known time domain filtering of prior art and the known position of airspace filter device in image encoding concern block diagram.Be that present technique field ordinary skill is known, be not described herein at this.
Among Fig. 3, " input picture " can be from any sample devices, " estimation " occurs in " input picture " and is stored between the coding and rebuilding image in " frame is deposited ", the motion match image of " input picture " and coding and rebuilding image is delivered to image pre-treatment filtering together, estimation matching image in filtering result and the coding and rebuilding carries out difference and calculates, the residual signals that obtains is by " DCT ", " quantification " carries out " variable-length encoding " afterwards, the result of variable-length encoding can send or store, signal before the variable-length encoding simultaneously is by " inverse quantization ", " anti-DCT ", carry out coding and rebuilding, the result of reconstruction is stored in " frame is deposited ".
Fig. 4 has described an embodiment who realizes the image pre-treatment filter of this method, is image pre-treatment filter in the dotted line wherein.Concrete treatment step is:
The first step is the estimation that unit carries out vision signal with the piece.
Can adopt image code domain three-step approach commonly used, motion estimation search algorithms such as rhombus therapy are aim to find optimum match block fast, can obtain the position and the matching error SAD of best matching blocks by estimation
Min:
Wherein:
The motion vector of X:x direction;
The motion vector of Y:y direction;
X (i, j): the position is at i, the original image pixels value of j;
Pred (i+X, j+Y): the position is at i+X, the coding and rebuilding image pixel value of j+Y;
Pred (i+X, j+Y): the position is at i+X, the coding and rebuilding image pixel value of j+Y;
SAD
Min: the matching error of best matching blocks;
Second step was determined the filtering characteristic of AR low pass filter according to the matching error of estimation match block:
a＝CLIP(C*SAD
min，0，1)
Wherein:
A: follow-up AR filter coefficient, scope is between 0 and 1;
C: be the constant relevant with the size of piece;
In the method that the present invention adopts, brightness and chrominance block are handled respectively, promptly luminance block is according to the matching error of luminance block, and chrominance block is calculated the AR filter coefficient respectively according to the matching error of chrominance block, can reduce the hangover of chrominance block like this.
The 3rd step was carried out time-domain filtering to " block to be encoded "
Adopt the filter coefficient of determining in second step, all pixels in the module unit carried out AR filtering:
Wherein:
Corresponding residual signals to be encoded is:
Follow-up cataloged procedure is identical with general cataloged procedure.
Can think; the foregoing description only is described in the mode of enumerating; it is conspicuous that many various changes and changes of making according to method of the present invention can also be arranged to one skilled in the art, can not break away from the protection range of design of the present invention and additional claim.
Claims (6)
1. image preprocessing method that is used for video signal compression is characterized in that: a. combines estimation and general time-domain filtering; B. be that the motion match piece of unit subregion and coding and rebuilding image carries out the time domain low-pass filtering with the piece to original image; C. adjust the filtering strength in different images zone according to the matching error self adaptation of estimation match block; D. image pre-treatment process is finished in the image encoding process simultaneously.
2. by the described image preprocessing method of claim 1: it is characterized in that described filtering carries out between original image and previous frame coding and rebuilding image.
3. by the described image preprocessing method of claim 1: it is characterized in that described filtering is to carry out between the motion match piece of original image and coding and rebuilding image.
4. by claim 1 or 2 or 3 described image preprocessing methods: the low pass filter cut-off frequency that it is characterized in that described different images piece is relevant with matching error.
5. by the described image preprocessing method of claim 4, it is characterized in that described matching error according to the estimation match block determines the low pass filter filtering characteristic: to luminance component is to determine the low pass filter cut-off frequency according to the matching error of brightness match block; Chromatic component is determined the low pass filter cut-off frequency according to the matching error of chromaticity match piece.
6. by the described image preprocessing method of claim 1: it is characterized in that it comprises the following step: 1. be the estimation that unit carries out vision signal with the piece; 2. determine the filtering characteristic of low pass filter according to the matching error of estimation match block; 3. " block to be encoded " carried out time-domain filtering.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA031533450A CN1514649A (en) | 2003-08-11 | 2003-08-11 | Image pretreatment method used for video frequency signal compression |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA031533450A CN1514649A (en) | 2003-08-11 | 2003-08-11 | Image pretreatment method used for video frequency signal compression |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1514649A true CN1514649A (en) | 2004-07-21 |
Family
ID=34240754
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA031533450A Pending CN1514649A (en) | 2003-08-11 | 2003-08-11 | Image pretreatment method used for video frequency signal compression |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1514649A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100414962C (en) * | 2005-04-18 | 2008-08-27 | 索尼株式会社 | Image signal processing device, camera system and image signal processing method |
| WO2012145989A1 (en) * | 2011-04-29 | 2012-11-01 | 杭州海康威视软件有限公司 | Cascade filter and method for dynamically setting calibrated denoising strength thereof |
| CN105791831A (en) * | 2009-07-07 | 2016-07-20 | 汤姆森特许公司 | Method and apparatus for using cooperative partition coding for region-based filters |
| CN105872538A (en) * | 2016-04-18 | 2016-08-17 | 广东中星电子有限公司 | Time-domain filtering method and time-domain filtering device |
| CN112509146A (en) * | 2020-11-23 | 2021-03-16 | 歌尔光学科技有限公司 | Image processing method, image processing device, electronic equipment and storage medium |
-
2003
- 2003-08-11 CN CNA031533450A patent/CN1514649A/en active Pending
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100414962C (en) * | 2005-04-18 | 2008-08-27 | 索尼株式会社 | Image signal processing device, camera system and image signal processing method |
| CN105791831A (en) * | 2009-07-07 | 2016-07-20 | 汤姆森特许公司 | Method and apparatus for using cooperative partition coding for region-based filters |
| US10708590B2 (en) | 2009-07-07 | 2020-07-07 | Interdigital Vc Holdings, Inc. | Methods and apparatus for collaborative partition coding for region based filters |
| US11483556B2 (en) | 2009-07-07 | 2022-10-25 | Interdigital Vc Holdings, Inc. | Methods and apparatus for collaborative partition coding for region based filters |
| US11895296B2 (en) | 2009-07-07 | 2024-02-06 | Interdigital Vc Holdings, Inc. | Methods and apparatus for collaborative partition coding for region based filters |
| US12375654B2 (en) | 2009-07-07 | 2025-07-29 | Interdigital Vc Holdings, Inc. | Methods and apparatus for collaborative partition coding for region based filters |
| WO2012145989A1 (en) * | 2011-04-29 | 2012-11-01 | 杭州海康威视软件有限公司 | Cascade filter and method for dynamically setting calibrated denoising strength thereof |
| CN105872538A (en) * | 2016-04-18 | 2016-08-17 | 广东中星电子有限公司 | Time-domain filtering method and time-domain filtering device |
| CN105872538B (en) * | 2016-04-18 | 2020-12-29 | 广东中星微电子有限公司 | Time domain filtering method and time domain filtering device |
| CN112509146A (en) * | 2020-11-23 | 2021-03-16 | 歌尔光学科技有限公司 | Image processing method, image processing device, electronic equipment and storage medium |
| CN112509146B (en) * | 2020-11-23 | 2023-06-20 | 歌尔科技有限公司 | Image processing method, device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1156168C (en) | Quick video motion estimating method | |
| CN1110962C (en) | Adaptive quantizer for use in video signal encoding system | |
| CN1098597C (en) | Noise estimation and reduction apparatus for video signal processing | |
| CN1201590C (en) | Video coding method using block matching process | |
| CN1184820C (en) | Coding and Noise Filtering Image Sequences | |
| CN101040532A (en) | Deblocking filter | |
| CN1968419A (en) | Image encoding method and apparatus and image decoding method and apparatus using characteristics of the human visual system | |
| US20030035586A1 (en) | Decoding compressed image data | |
| CN1347621A (en) | Reducing 'blocking picture' effects | |
| CN1520191A (en) | Method and device for removing blocking artifacts of MPEG signals | |
| CN1372224A (en) | Method for image predenoising | |
| CN1535019A (en) | A loop filtering method and loop filter | |
| CN100490537C (en) | Method for reducing image blocking effect | |
| CN1514649A (en) | Image pretreatment method used for video frequency signal compression | |
| CN1139255C (en) | Improved post-processing method and device for image signal decoding system | |
| CN1758759A (en) | Switching detection method of video scene | |
| CN1969556A (en) | Method for chroma deblocking | |
| CN1968415A (en) | Processing of compressed video signal | |
| CN1665299A (en) | Method for designing architecture of scalable video coder decoder | |
| Abhayaratne et al. | Embedded-to-lossless coding of motion-compensated prediction residuals in lossless video coding | |
| US20150304686A1 (en) | Systems and methods for improving quality of color video streams | |
| CN1770880A (en) | Time-domain filtering method for video code | |
| CN1668106A (en) | Method and device for removing block effect of image frame by loop filtering | |
| KR100885441B1 (en) | Filtering method for block boundary region | |
| CN100490534C (en) | Continuous image compression method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |