CN100574440C - A kind of image and video anti-error code method based on pixel rearrangement - Google Patents
A kind of image and video anti-error code method based on pixel rearrangement Download PDFInfo
- Publication number
- CN100574440C CN100574440C CN 200710168411 CN200710168411A CN100574440C CN 100574440 C CN100574440 C CN 100574440C CN 200710168411 CN200710168411 CN 200710168411 CN 200710168411 A CN200710168411 A CN 200710168411A CN 100574440 C CN100574440 C CN 100574440C
- Authority
- CN
- China
- Prior art keywords
- image
- images
- error code
- pixel
- subimage
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The present invention relates to a kind of image and video anti-error code method based on pixel rearrangement, the method that is adopted is: before encoder is to two field picture coding, carry out following steps: 1. according to pixel column image is carried out 1/N and extract, the subimage that to obtain N reduced height be 1/N; 2. these subimages are spliced into the new images of sizes such as and original image in vertical direction, this new images is obtained a frame decoding image through decoder decode after, carry out following steps: 3. recover the N number of sub images from decoded picture; 4. with a plurality of subimages weave in line by line successively, recover original image; 5., then do spatial prediction and replace the image of losing with adjacent sub-images if there is error code to cause parts of images to decode.Thought of the present invention is the method with similar holographic, by pixel rearrangement to image, make the part of image have the ability of representing entire image to a certain extent, thereby when error code took place, the enough remaining images of energy recovered the part of missing image preferably.
Description
Technical field
The invention belongs to the processing and the transmission field of video and picture signal.A kind of specifically image and video anti-error code method based on pixel rearrangement.
Background technology
Along with developing rapidly of video coding technique, network technology and memory technology, all obtained application more and more widely, for example video conference, video monitoring, DVD optic disk system and digital photo album etc. based on the transmission and the storage system of video and image compression encoding.Yet any communication network and medium can not guarantee that all data are reliable fully.And various video compressing and encoding methods commonly used have all adopted traditional hybrid video coding method based on time prediction basically, though obtained very high compression ratio, also responsive extraordinarily to error code.This has just limited video system in the channel of some high bit errors and the application on the medium.Same situation also occurs in the image compression encoding application.Therefore, how to make video coding algorithm have good error recovery capabilities, it is used, particularly significant to Internet video and the wireless video that is subjected to extensive concern at present.
In existing video coding, because the influence of time prediction, if mistake appears in a certain frame video data, not only mistake can appear in the decoded picture of this frame, even the subsequent video frame data are all correct, can cause also thereafter that mistake appears in the decoded picture of some frames, till next I frame (non-predictive frames).Therefore, in the video signal decoder, in order to obtain better visual performance, if found error code, before next I frame, if decoder does not freeze to export the parts of images that picture just can only be lost by the whole bag of tricks conjecture as much as possible, this is called as error concealment.In having the video communication system of feedback channel, though can notify encoder to upgrade I frame or similar method recovery image by decoder, but because the round-trip delay of network, before the I frame arrived, decoder also only freezed to export picture and recovers two selections of image by error concealment as far as possible.So the image quality the when quality of error concealment method meets with error code to decoder is most important.
For video coding, common error concealment method generally can be classified as two classes: 1, time prediction method: generally use the motion vector of the estimation of motion vectors erroneous macroblock of adjacent macroblocks, the reference block with the motion vector appointment directly substitutes errored block again; 2, spatial prediction method: recover the macro block lost with the gradient information of adjacent macroblocks and neighbor by the method for interpolation or filtering.In actual applications, these two kinds of methods can both obtain certain effect under given conditions.Adopt the time prediction method, recover the size of sum of errors coded residual component that quality depends on the motion vector of recovery, uncomplicated for image motion, recovery effects preferably can be obtained in the ground that coded residual is little.Adopt the spatial prediction method, generally can only recover the low frequency composition of image, the image detail that shows as the high frequency composition is difficult to recover.For image encoding, life period is not predicted, can only adopt the method for spatial prediction, the situation all fours.
But, no matter be the time prediction method, or the spatial prediction method, any is arranged is identical, that is exactly all to need to utilize the data of adjacent macroblocks to carry out error concealment.If the data of adjacent macroblocks also are wrong, then error concealment failure.Time prediction method and spatial prediction method all are base unit with the macro block, and the time prediction method generally need be lost the motion vector of macro block by the estimation of motion vectors of adjacent macroblocks, and the spatial prediction method need recover to lose macro block with neighbor.That is to say, it can not be continuous that existing error concealment technology generally requires wrong macro block, and in fact the image/video data often show as the burst error of continuous a plurality of macro blocks, and this has had a strong impact on the error concealment effect of existing time prediction method and spatial prediction method.
Summary of the invention
The objective of the invention is to propose a kind of simple effectively anti-error code method, wrong continuously image is appearred in data still have good error concealment ability, handle simple, be easy to realize with DSP or hardware, do not need existing encryption algorithm is made any modification, to the same image and the video anti-error code method that is suitable for of image, to overcome the existing error concealment method deficiency that effect is degenerated under the continuous error situation of view data based on pixel rearrangement with video.
To achieve these goals, the method applied in the present invention is:
Before encoder is to two field picture coding, carry out following steps:
First step, according to pixel column image is carried out 1/N and extract, the subimage that to obtain N reduced height be 1/N;
Second step, these subimages are spliced into the new images of sizes such as and original image in vertical direction, this new images is encoded, new images behind this coding is offered decoder through transmission channel or medium, after the process decoder decode obtains a frame decoding image, carry out following steps:
Third step, recover the N number of sub images from decoded picture;
The 4th step, with a plurality of subimages weave in line by line successively, recover original image;
If the 5th step has error code to cause parts of images to decode, then do spatial prediction and replace the image of losing with adjacent sub-images;
The 6th step, for video decode, if error code has taken place, also to extract the image that the 5th step obtains according to first and second steps and splice the back again and replace the reference frame that decoded picture saves as decoder.
The abstracting method of above-mentioned first step be with the row of pixel column number to the N delivery, the row that mould is identical is drawn into together by former order, constitutes a number of sub images, obtains the N number of sub images altogether, the mould value is the sequence number of subimage.
The joining method of above-mentioned second step is that the N number of sub images is stored together from the top down continuously according to the sequence number incremental order.
The method of the recovery subimage of above-mentioned third step is the inverse process of second step.
The method of the recovery original image of above-mentioned the 4th step is the inverse process of first step.
The concrete grammar of above-mentioned the 5th step is: for any zone of image, because adjacent N is capable to be dispersed in the N number of sub images, be evenly distributed in the entire image, as long as the counterpart of this zone in the N number of sub images all do not lost, so in the input picture of the 5th step, (M＜N) individual continuous erroneous pixel row always is clipped between two correct pixel columns M, recovers M continuous erroneous pixel row by utilize spatial prediction between correct pixel column arbitrarily.
The method of above-mentioned the 6th step is for video decode, and the image that the 5th step is obtained extracts according to first and second steps again and splices the back and replaces the reference frame that decoded picture saves as decoder.Be used to improve the quality of subsequent frame time reference, improve decoding effect.
Thought of the present invention is the method with similar holographic, by pixel rearrangement to image, make the part of image have the ability of representing entire image to a certain extent, thereby when error code took place, the enough remaining images of energy recovered the part of missing image preferably.Experiment shows that method of the present invention has very strong error correcting ability, and it is example that the image that causes with error burst is lost continuously, as long as the normal part of image surpasses 1/N, still can recover entire image.Because the correlation of image adjacent lines is very high, is not more than at 3 o'clock at N, image restored quality of the present invention is very high, and when the picture large tracts of land was lost, its visual effect obviously was better than other method.The present invention can be used for anti-error code video and image encoding, and is without any special requirement, compatible good to encoder, is a kind of method in common.
Description of drawings
Fig. 1 is a video coding/decoding system block diagram of the present invention.
Fig. 2 is example with N=3 for the embodiment of the invention, extracts subimage 0 and obtains subimage 1 and subimage 2.
Fig. 3 is example with N=3 for the embodiment of the invention, and with subimage 0, subimage 1 and subimage 2 are spliced into new images.
Fig. 4 is example with N=3 for the embodiment of the invention, and with subimage 0, subimage 1 and subimage 2 recover to come out from decoded picture.
Fig. 5 is example with N=3 for the embodiment of the invention, and with subimage 0, subimage 1 and subimage 2 recover source images.
Fig. 6 be the embodiment of the invention with N=2, image is the CIF size, error code causes the 8th, 9 macro-block line can't be decoded as example, and the restoration methods of losing image is described, the part of grey is that error code causes the image lost among the figure.
Fig. 7 is example with N=2 for the embodiment of the invention, approach based on linear interpolation figure when continuous 1 row is lost.
Fig. 8 be the embodiment of the invention with N=3, image is the CIF size, error code causes the 6th to 12 macro-block line can't be decoded as example, and the restoration methods of losing image is described, the part of grey is that error code causes the image lost among the figure.
Fig. 9 is example with N=3 for the embodiment of the invention, approach based on linear interpolation figure when continuous 2 row are lost.
Embodiment
The present invention is described in further detail below in conjunction with drawings and Examples.
Principle of the present invention is: the pixel column of image rearranges, adjacent pixels row in the original image is extracted the number of sub images into N line by line, each subimage all can be considered an entire image sub-sampling in vertical direction and represents like this, can represent entire image separately to a certain extent.Again the N number of sub images is stitched together to form in vertical direction in order and resets image, thereby make original image on the sub-sampling basis, have N expression, formed the redundancy on the sub-sampling basis.If wherein the experimental process image is destroyed,, just can recover entire image to a certain extent as long as wherein any one subimage exists.By resetting, make that the adjacent pixels row is drawn in the different subimages in the original image, in resetting image, leave distance place far away in, thereby be not easy to be destroyed simultaneously by burst error.Therefore,, cause the part macro block to decode, also only can influence wherein some subimage even continuous mistake appears in certain part in the view data.As long as N number of sub images corresponding position is not all lost, because the high correlation of image adjacent lines still can be done the image that the spatial prediction recovery is lost with remaining intact subimage between neighborhood pixels is capable.So when getting the N number of sub images, lose continuously under the situation that is no more than (N-1)/N at image and still can recover original image.Under most situations, the error in data on transmission medium and the recording medium all is burst and distributes, so mostly the missing image that error code causes is continuous mistake also.Experiment shows that under the wrong continuously situation less than (N-1)/N of image, method of the present invention still can be recovered original image, is not more than at 3 o'clock at N, and image restored peak value quality and visual effect obviously are better than other method.Because method of the present invention does not relate to coding, the decode procedure of video and image, and only doing mistake in image inside covers, so the present invention is general to video coding and image encoding, can cooperate any video coding algorithm and image encoding algorithm, be the very high method of a kind of versatility.Owing to adopt the processing based on pixel column, simple and concurrency is high, the present invention is easy to realize with DSP or hardware.
Concrete grammar of the present invention is: did pre-treatment (Fig. 1) before encoder is to two field picture coding earlier:
The first step, source images is carried out 1/N according to pixel column extract, obtain the subimage that N vertical height is reduced into 1/N: the pixel column of source images number is for since 0 progressive whole number successively, will the number of going to the N delivery, the result is 0 to N-1, all pixel columns that delivery is come to the same thing store together according to former order, obtain the N number of sub images.Because picturedeep is the integral multiple of N not necessarily, so the height of subimage has 1 difference.
With N=3 is example, extracts subimage 0, subimage 1 and subimage 2 (Fig. 2).
Second step, subimage is spliced into the new images of sizes such as and original image in vertical direction: the N number of sub images is stitched together in the vertical direction byte according to sequence number.With N=3 is example, and with subimage 0, subimage 1 and subimage 2 are spliced into new images (Fig. 3).
After obtaining a frame decoding image, does decoder reprocessing:
The 3rd step, recover the N number of sub images from decoded picture: this is the inverse process in second step, is example with N=3, and with subimage 0, subimage 1 and subimage 2 recover to come out (Fig. 4) from decoded picture.
The 4th goes on foot, a plurality of subimages is spliced successively line by line, and recover original image: this is the inverse process of the first step, is example with N=3, and with subimage 0, subimage 1 and subimage 2 recover source images (Fig. 5).
If the 5th step had error code to cause parts of images to decode, then do spatial prediction and replace the image of losing with adjacent sub-images, this is a core of the present invention: by resetting, the adjacent pixels row is drawn in the different subimages in the original image, thereby is not easy to be destroyed simultaneously by burst error.Therefore,, cause the part macro block to decode, also only can influence wherein some subimage even continuous mistake appears in certain part in the view data.As long as N number of sub images corresponding position is not all lost, because the high correlation of image adjacent lines still can be done the image that the spatial prediction recovery is lost with remaining intact subimage between neighborhood pixels is capable.So when getting the N number of sub images, lose continuously under the situation that is no more than (N-1)/N at image and still can recover original image.
The 6th step, for video decode, if error code has taken place, also to extract the image that the 5th step obtains according to first and second steps and splice the back again and replace decoded picture and save as reference frame.Be used to improve the quality of subsequent frame time reference, improve decoding effect.
Space predicting method with the most simple and practical linear interpolation is an example below, and the restoration methods of error image is described.
For the image of losing continuously,,, can guarantee promptly behind the 4th step splicing subimage that always the pixel column of losing arbitrarily is effectively between the pixel column to any N as long as the image of losing is no more than entire image (N-1)/N.Suppose to have between capable a of valid pixel and the b the individual pixel column of losing of M, then as figure below, to the capable pixel column of losing of m, the general formula of linear interpolation is as follows:
xm
i＝((M-m-1)a
i+(m+1)b
i+M/2)/M
The ordinate of subscript i remarked pixel wherein.
Respectively with N=2 and N=3, image is that the CIF size is an example, and recovery ten thousand methods of losing image are described below.
During N=2, suppose that error code causes decoder can't decode the 8th, 9 macro-block line, i.e. the 128th to 159 pixel column (the grey color part of Fig. 6).As can be seen, behind the 4th EOS, all the pixel provisional capitals that can't decode are sandwiched between the normal pixel row among the figure.The n that loses is capable can be recovered by the linear interpolation that n-1 is capable and n+1 is capable, the n here is 1,3,5 ..., 31 and 256,258,260 ..., 286.
Approach based on linear interpolation such as Fig. 7, formula is as follows:
x
i＝(a
i+b
i+1)/2
During N=3, suppose that error code causes decoder can't decode the 6th to 12 macro-block line (subimage 1 all with the 1st macro-block line of subimage 2), i.e. the 96th to 207 pixel column (the grey color part of Fig. 7).As can be seen, behind the 4th EOS, all pixel columns that can't decode are still separated by the normal pixel row among Fig. 8.The image of losing can be divided into two parts, and n, the n+1 that loses is capable can be recovered by the linear interpolation that n-1 is capable and n+2 is capable, the n here is 1,4,7 ..., 46.
Approach based on linear interpolation such as Fig. 9, formula is as follows:
x0
i＝(2a
i+b
i+2)/3
x1
i＝(a
i+2b
i+2)/3
Similar with N=2, the n that loses is capable can be recovered by the linear interpolation that n-1 is capable and n+1 is capable, the n here is 49,52,55 ..., 286.The situation of this part approach based on linear interpolation and N=2 is identical.
Because above each step all is with behavior unit to treatment of picture, parallel carrying out between a plurality of pixels in being expert at so be very beneficial for not only that the SIMD instruction set with CPU realizes on PC, also is highly susceptible to realizing with DSP or hardware.In addition, method of the present invention is only done mistake at image inner utilization spatial prediction and is covered, the coding, the decode procedure that do not relate to video and image so the present invention is general to video coding and image encoding, can cooperate any video coding algorithm and image encoding algorithm.This is the very high method of a kind of practicality.
The content that is not described in detail in this specification belongs to this area professional and technical personnel's known prior art.
Claims (6)
1, a kind of image and video anti-error code method based on pixel rearrangement comprise following steps:
Before encoder is to two field picture coding, carry out following steps:
First step, according to pixel column image is carried out 1/N and extract, the subimage that to obtain N reduced height be 1/N;
Second step, these subimages are spliced into the new images of sizes such as and original image in vertical direction, this new images is encoded, new images behind this coding is offered decoder through transmission channel or medium, after the process decoder decode obtains a frame decoding image, carry out following steps:
Third step, recover the N number of sub images from decoded picture;
The 4th step, with a plurality of subimages weave in line by line successively, recover original image;
If the 5th step has error code to cause parts of images to decode, then do spatial prediction and replace the image of losing with adjacent sub-images;
The 6th step, for video decode, if error code has taken place, the image that the 5th step is obtained extracts according to first and second steps again and splices the back and replaces the reference frame that decoded picture saves as decoder.
2, image and video anti-error code method based on pixel rearrangement as claimed in claim 1, it is characterized in that: the abstracting method of first step is number to the N delivery with the row of pixel column, the row that mould is identical is drawn into together by former order, constitute a number of sub images, obtain the N number of sub images altogether, the mould value is the sequence number of subimage.
3, image and video anti-error code method based on pixel rearrangement as claimed in claim 1, it is characterized in that: the joining method of second step is that the N number of sub images is stored together from the top down continuously according to the sequence number incremental order.
4, image and video anti-error code method based on pixel rearrangement as claimed in claim 1, it is characterized in that: the method for the recovery subimage of third step is the inverse process of second step.
5, image and video anti-error code method based on pixel rearrangement as claimed in claim 1, it is characterized in that: the method for the recovery original image of the 4th step is the inverse process of first step.
6, image and video anti-error code method based on pixel rearrangement as claimed in claim 1, it is characterized in that: the concrete grammar of the 5th step is: for any zone of image, because adjacent N is capable to be dispersed in the N number of sub images, be evenly distributed in the entire image, as long as the counterpart of this zone in the N number of sub images all do not lost, so in the input picture of the 5th step, M continuous erroneous pixel row always is clipped between two correct pixel columns arbitrarily, by between correct pixel column, utilizing spatial prediction to recover M continuous erroneous pixel row, wherein M＜N.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200710168411 CN100574440C (en) | 2007-11-22 | 2007-11-22 | A kind of image and video anti-error code method based on pixel rearrangement |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200710168411 CN100574440C (en) | 2007-11-22 | 2007-11-22 | A kind of image and video anti-error code method based on pixel rearrangement |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101267555A CN101267555A (en) | 2008-09-17 |
| CN100574440C true CN100574440C (en) | 2009-12-23 |
Family
ID=39989638
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200710168411 Expired - Fee Related CN100574440C (en) | 2007-11-22 | 2007-11-22 | A kind of image and video anti-error code method based on pixel rearrangement |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100574440C (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111683216A (en) * | 2020-05-25 | 2020-09-18 | 福建云视听信息科技有限公司 | Interlaced video processing method, system, device and medium in interlacing mode |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102098500A (en) * | 2011-03-02 | 2011-06-15 | 天津大学 | Error collecting method for improving transmission performance of eight-view auto-stereoscopic video network |
| CN102164286B (en) * | 2011-05-27 | 2012-09-26 | 天津大学 | Method for hiding transmission error of eight-viewpoint free three-dimensional video |
| CN110769206B (en) * | 2019-11-19 | 2022-01-07 | 深圳开立生物医疗科技股份有限公司 | Electronic endoscope signal transmission method, device and system and electronic equipment |
-
2007
- 2007-11-22 CN CN 200710168411 patent/CN100574440C/en not_active Expired - Fee Related
Non-Patent Citations (2)
| Title |
|---|
| 全景图拼接算法的设计与实现. 杨刚.重庆工学院学报（自然科学版），第21卷第9期. 2007 |
| 全景图拼接算法的设计与实现. 杨刚.重庆工学院学报（自然科学版），第21卷第9期. 2007 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111683216A (en) * | 2020-05-25 | 2020-09-18 | 福建云视听信息科技有限公司 | Interlaced video processing method, system, device and medium in interlacing mode |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101267555A (en) | 2008-09-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101055738B1 (en) | Method and apparatus for encoding/decoding video signal using prediction information of intra-mode macro blocks of base layer | |
| KR100667806B1 (en) | Method and apparatus for video encoding and decoding | |
| JP5918328B2 (en) | Image encoding method and image decoding method | |
| KR20010076708A (en) | Transmission/receipt data processing method for inhibiting error propagation in digital image data communication system and computer-readable medium therefor | |
| JP2012502592A (en) | Decoding system and method using parallel processing | |
| US20090060049A1 (en) | Method and system for calculating flag parameter of image block | |
| CN113766249B (en) | Loop filtering method, device, equipment and storage medium in video coding and decoding | |
| CN102204242A (en) | Method and system for increasing frame-display rate | |
| CN108965887B (en) | Video information hiding method and device based on inter-block decoupling | |
| CN100574440C (en) | A kind of image and video anti-error code method based on pixel rearrangement | |
| JP3351645B2 (en) | Video coding method | |
| CN101291436B (en) | Video coding/decoding method and device thereof | |
| CN101163250B (en) | Boundary gradient based video stream fault tolerance method | |
| KR20060063608A (en) | Method and apparatus for conducting residual prediction on a macro block when encoding/decoding video signal | |
| CN103997651A (en) | Data compression method and device for composite images | |
| Song et al. | A data embedding scheme for H. 263 compatible video coding | |
| CN105430406A (en) | Distributed video coding framework | |
| CN1147157C (en) | Decoding device and method | |
| KR100883591B1 (en) | Method and apparatus for encoding/decoding video signal using prediction information of intra-mode macro blocks of base layer | |
| CN101601297B (en) | Method and apparatus for encoding groups of pictures of a video sequence | |
| CN102300098A (en) | Video fault tolerance coding method based on reversible information hiding and redundancy technology | |
| CN111212288B (en) | Video data encoding and decoding method and device, computer equipment and storage medium | |
| Lin et al. | A novel error resilience using reversible data embedding in H. 264/AVC | |
| Lin et al. | An error resilient technique using reversible data embedding in H. 264/AVC | |
| US20080049836A1 (en) | Method and System for a Fast Video Transcoder |
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 | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20091223 Termination date: 20111122 |