CN100481954C - System and method for motion compensating - Google Patents
System and method for motion compensating Download PDFInfo
- Publication number
- CN100481954C CN100481954C CN 200510123882 CN200510123882A CN100481954C CN 100481954 C CN100481954 C CN 100481954C CN 200510123882 CN200510123882 CN 200510123882 CN 200510123882 A CN200510123882 A CN 200510123882A CN 100481954 C CN100481954 C CN 100481954C
- Authority
- CN
- China
- Prior art keywords
- frame image
- module
- reference frame
- discrete cosine
- cosine transform
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims description 11
- 238000007906 compression Methods 0.000 claims abstract description 32
- 230000006835 compression Effects 0.000 claims abstract description 30
- 230000006837 decompression Effects 0.000 claims abstract description 20
- 238000013139 quantization Methods 0.000 claims description 11
- 230000005540 biological transmission Effects 0.000 description 5
- 238000006243 chemical reaction Methods 0.000 description 3
- 239000013256 coordination polymer Substances 0.000 description 1
- 238000013144 data compression Methods 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 230000002427 irreversible effect Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
This invention provides a motion compensation system for realizing the motion compensation to digital images of high resolution based on limited storage module space including a compression/decompression module to compress the reference frame image data necessary for the motion compensation to reduce the request to the size of a storage module in the system.
Description
Technical field
The present invention relates to digital image processing field, particularly the motion compensation field.
Technical background
Motion Picture Experts Group (MPEG) standard because of its multi-level picture quality and transmission speed, have gradability, flexibility and adaptability widely, become the recognised standard of DVB, HDTV source encoding.In recent years, except that transmission, in collection, making and the broadcast system of television operating center and in the multimedia field, also obtained using widely.
For the ease of random access and editor to image sequence, the structure that MPEG has stipulated layering to video data stream as shown in Figure 1, begin to be followed successively by from top layer: ground floor is the image sequence layer, and it comprises the data and the sequence full stop of sequence head, several image sets layers.The second layer is image sets (GOP, Group ofPicture), and it is made up of one or more groups intracoded frame or the non-intracoded frame image of definition.Every group comprises group head and image layer data.The 3rd layer is image layer itself, is made up of image head and macro block bar layer data.The 4th layer is macro block bar layer, and it is made up of bar layer head and several continuous macro blocks.Layer 5 is a macroblock layer, is made up of macro block head and piece layer data, and image is a benchmark with the brightness data matrix, is divided into the macro block of 16 * 16 pixels, as the elementary cell of motion compensation.Macro block is divided into luminance macroblock and chrominance macroblock, and the chrominance macroblock size is relevant with the sampling form.Layer 6 is that the piece layer is made up of view data and end of block character, is the unit that carries out discrete cosine transform.Comprised decode required initial code, regularly and other relevant parameter and information of MPEG code stream at the corresponding sequence head of this six layer data structures, image sets head, image head, macro block bar head, the first-class head of macro block.
The vision signal intraframe coding is to utilize the spatial coherence of signal to realize Information Compression, utilizes discrete cosine transform coding technology in the frame, and the spatial coherence of excavating between the same picture neighborhood pixels is realized the active data compression.Interframe encode then is to utilize the correlation of picture signal on time shaft to realize Information Compression.Statistical measurement shows, when scenery does not contain strenuous exercise, occurrence scene does not switch and video camera when not doing obvious motion, and the pixel of the locus correspondence between consecutive frame has stronger correlation than neighbor in the frame.
As previously mentioned, comprise three types picture frame in each image sets: I frame, P frame, B frame.The I frame is only done intraframe coding, the compensation interframe of not taking exercises prediction.Each image sets begins with an I frame, is convenient to receive at random and regularly carries out frame and refresh, to prevent transmission error code because inter prediction may cause long-time diffusion.The P frame is a forward predicted frame, by the compensation prediction of taking exercises at the I of preceding transmission frame or P frame (reference frame).The B frame is a bi-directional predicted frames, its simultaneously by transmitted in image sequence, be in before this frame and two reference frames behind this frame are made forward motion compensation prediction and consequent motion compensated prediction respectively.The B frame can not be used as the reference frame that other frame carries out motion compensated prediction.
So-called motion can be understood as the vector value of a two dimension, according to this vector value and in conjunction with front/rear information with reference to intra-frame macro block, can obtain corresponding macro block information in this two field picture.Motion compensating system has improved the coding of P frame and B frame by removing the temporal redundancy of interframe.
Yet motion compensating system need possess certain memory space and be used for the view data of storage of reference frames.Along with improving constantly of digital video signal resolution, also just constantly improve for the size of memory space described in the motion compensating system and the requirement of bandwidth, just improve the cost of video decoding system so virtually greatly, brought great puzzlement for the manufacturer that makes video receiving apparatus (as digital TV set-top box).
Summary of the invention
The method that the purpose of this invention is to provide a kind of motion compensating system and carry out motion compensation ensures and carries out on the basis of motion compensation at the image that does not influence high-res, reduces the requirement to memory module size in the motion compensating system.For achieving the above object, the present invention has put down in writing following technical scheme:
A kind of motion compensating system comprises an analysis module, an inverse discrete cosine transform module, a motion compensating module, memory module, and a compression and a decompression module.Described analysis module receives image sequence and analyzes, and described motion compensating module is judged this two field picture type of coding and described memory module is carried out read/write operation.When carrying out read operation, described decompression module decompresses to the reference frame image data that come from the described memory module; When carrying out write operation, described motion compensating module is to the data that comprise motion vector information that come from described inverse discrete cosine transform module and describedly carry out combination through the reference frame image data that decompress and obtain this frame image data and export; If this frame is independent image frame or forward predicted frame, then this frame image data is compressed and write described memory module through described compression module.
Motion compensating module also comprises motion compensated predictor described in the described motion compensating system, read command register, write command register, motion compensation read through model and motion compensation writing module.Described motion compensated predictor is sent corresponding read/write command respectively to described read/write command register and this two field picture type of coding is judged described motion compensation read/write modules receives corresponding read/write command respectively and described memory module is carried out read/write from described read/write command register according to the data that receive from described analysis module.
The motion compensation read through model comprises read operation address generator, adjuster, half pixel digital difference detector, read buffer register described in the described motion compensating system; Described motion compensation writing module comprises the write address generator, and colligator is write buffer register.When carrying out read operation, the described address generator of reading produces a certain read operation address according to the order that comes from described read command register, be sent to described decompression module corresponding to the reference frame image data of this read operation address in the described memory module and carry out decompression operation, the reference frame image data after decompressing are successively through being sent in the described read buffer register after described adjuster, the half pixel digital difference detector; When carrying out write operation, described colligator carries out combination to data that comprise motion vector information that come from described inverse discrete cosine transform module and the reference frame image data through decompressing that come from described read buffer register and obtains this frame image data and output; Judge through described motion compensated predictor, if this two field picture is independent frame or forward predicted frame, then this frame image data that generates after combination is sent to described writing in the buffer register through after the compression of described compression module, described write address generator produces a certain write address according to the order that comes from described write command register, and the described compressed view data of writing in the buffer register is stored in the described memory module position that should write address.
In addition, comprise the device that the view data after the described combination is carried out discrete cosine transform (DCT) described in the described motion compensating system in the compression module; To the device that quantizes through the resulting coefficient of described discrete cosine transform; And the device that the discrete cosine transform coefficient after quantizing is compressed.Decompression module described in the described motion compensating system comprises the device that described reference frame image data are decompressed; Reference frame image data after decompressing are carried out the device of de-quantization; And to carried out the device of inverse discrete cosine transform (IDCT) by the reference frame image data of de-quantization.In described compression module, can also comprise that between described discrete cosine conversion device and described quantization device one is used for device that the ac coefficient of discrete cosine transform coefficient is expanded; Between described quantization device and described compression set, can also comprise a clamp device.In described decompression module, can also comprise that between described de-quantization device and described inverse discrete cosine transform device one is used for the ac coefficient of discrete cosine transform coefficient is carried out the device of contrary expansion.
In addition, the present invention has also write down a kind of method of carrying out motion compensation, and this method comprises: the figure frame sequence that receives is carried out analysis and judgement, and produce the operation of read/write command; Read the address according to the read command generation, and read the operation that the address is read the reference frame image data and decompressed according to this; According to write order, carry out combination to described through reference frame image data that decompress and the data that comprise motion vector information through inverse discrete cosine transform, produce the operation of this frame image data and output; If this frame is independent image frame or forward predicted frame, then produce write address according to write order, described frame image data compressed and writes the operation of described write address.
Adopt the motion compensating system that the present invention put down in writing, under the prerequisite that does not change the view data throughput, by the reference frame image data in the operation of motion compensation are compressed, reduced the requirement of motion compensating system, thereby greatly reduced the cost of video receiving apparatus for the size and the bandwidth of its storage inside module.
Description of drawings
Fig. 1 is a MPEG video data stream structure chart;
Fig. 2 is the structure chart of motion compensating system among the present invention;
Fig. 3 A and Fig. 3 B are the working state figures of motion compensating system internal compression module among the present invention;
Fig. 4 A and Fig. 4 B are the working state figures of the inner decompression module of motion compensating system among the present invention.
Embodiment
The present invention provides a kind of motion compensating system, as shown in Figure 2 for realizing aforesaid goal of the invention.Video signal analysis module 100 is responsible for receiving the vision signal that comes from the video transmission device and it is analyzed, the view data that will comprise motion vector information sends to inverse discrete cosine transform module 200, and general's signal by analysis sends to the motion compensated predictor 310 in the motion compensating module 300 simultaneously.310 pairs of this two field picture type of codings of described motion compensated predictor are judged, if this two field picture then issues a signal to write command register 330 for not needing the I type frame with reference to other frame data.Motion compensation writing module 350 is received the write order from write command register 330, the transfer of data that then will come from inverse discrete cosine transform module 200 is to colligator 352, owing to do not need to carry out any binding operation, the data of this process inverse discrete cosine transform are output to the display system of video receiving apparatus as this frame image data, while described frame image data also is admitted to compression module 600 and compresses, data after compression are written into writes buffer register 353, and finally is written into memory module 400 conducts with reference to frame image data.
If described motion compensated predictor 310 detects this two field picture and is encoded to the P type frame that need carry out forward prediction, then send read command to read command register 320.Described read command register 320 indicates the forward frame view data of reading the required reference of address generator 341 generations to be arranged in the address of memory module 400 according to received order.Because stored in memory module 400 all is reference frame image data through overcompression, therefore ought describedly read can be sent to decompression module after the pairing forward reference frame view data in address is read out and decompress for 500 kinds.Adjust and interpolation by adjuster 342 and half pixel digital difference detector 343 successively through the forward reference frame view data after decompressing, be sent to again in the read buffer register 344 and await orders.Described motion compensated predictor 310 is sent write order to described write command register 330.Write command register 330 indication write address generators 351 produce a write address, so that the data of this two field picture are write the middle relevant position of memory module 400.When carrying out write operation, the data that comprise motion vector information that come from described inverse discrete cosine transform module 200 are sent in the colligator 352 and combine with the forward reference frame view data that comes from described read buffer register 344, thereby obtain the data of this two field picture and be output.At this moment a resulting frame data are uncompressed, therefore need bigger memory space to hold these data.Yet through after the compression of compression module 600, the occupied memory space of this frame image data is reduced greatly.Be sent to through the data of this two field picture of overcompression and write in the buffer register 353, and be written into memory module 400 as new reference frame image data according to the write address that described write address generator 351 is produced.
If described motion compensated predictor 310 detects this two field picture and is encoded to the Type B frame that need carry out two-way reference, then with behind the current B two field picture to I or P frame store in the described memory module 400 as the reference frame according to the flow process of aforementioned correspondence, and then carry out motion compensation at current B frame.When carrying out motion compensation at current B frame, described motion compensated predictor 310 sends data to read command register 320.Described read command register 320 according to received signal indication read address generator 341 and produce the forward direction of required reference and back to reference frame image data address of 400 in memory module.After described reference frame image data are read out, can be sent in the decompression module 500 and decompress.Adjust and interpolation by adjuster 342 and half pixel digital difference detector 343 successively through the reference frame image data after decompressing, be sent to again in the read buffer register 344 and await orders.Described motion compensated predictor 310 can be sent write order to described write command register 330.When carrying out write operation, the view data that comprises motion vector that comes from described inverse discrete cosine transform module 200 is sent in the colligator 352 and combines with the reference frame image data that come from described read buffer register 344, thereby obtains the data and the output of this two field picture.
Fig. 3 A, shown in the B is the application drawing that is positioned at the module 600 that motion compensating system compresses view data in the technical solution of the present invention, Fig. 4 A, shown in the B is the application drawing that is positioned at the module 500 that the reference frame image data is decompressed motion compensating system inside in the technical solution of the present invention.It all is data through overcompression that the existence of this compression/de-compression module makes the data of reading or write described memory module 400, thereby has reduced the requirement to the size and the bandwidth of described memory module.
As shown in Figure 3A, described compression module 600 comprises a discrete cosine transform (DCT) device D, quantization device Q and compression set P.Described discrete cosine conversion device D is to changing as piece of being made up of 8 pixels, and the conversion precision is 16, and resulting transformation parameter coefficient precision is 9.After carrying out discrete cosine transform, shown in Fig. 3 B, alternating-current parameter in the resulting discrete cosine transform parameter can also be through an expanding unit C, this expanding unit C expands described alternating-current parameter according to formula (1), and reference point breakpoint wherein is according to different alternating-current parameter value differences.
breakpoint+abs(coefficient-breakpoint)/2 (1)
Described parameter through discrete cosine transform is being quantized by a quantization device Q through the expansion back.After carrying out quantization operation, to carry out before the squeeze operation, described parameter can be carried out clamper according to the compressed format that will take to the parameter after quantizing by a clamp device CP.For example the compressed format that will take is " 87654... ", and on behalf of dc parameter, this take 8, is limited to 255 on the value; First alternating-current parameter takies 7, and span is-64 to+63; Other alternating-current parameters are followed the definite span separately of this compressed format.The parameter value that surpasses these spans will be by clamper, the just approximate immediate upper limit or the lower limit of getting.After carrying out clamper, described discrete cosine transform parameter is compressed with predetermined compressed format through a compression set P.
Fig. 4 A, B are depicted as the application drawing of described decompression module 500, and its operating process is the inverse process of described compression module 600, and only the clamper step in the compression process is irreversible step.
The above only is preferred embodiment of the present invention, and is in order to restriction the present invention, within the spirit and principles in the present invention not all, any modification of being done, is equal to replacement etc., all should be included within protection scope of the present invention.
Claims (5)
1. motion compensating system, comprise an analysis module, one inverse discrete cosine transform module, one motion compensating module and memory module, it is characterized in that described motion compensating module also comprises motion compensated predictor, colligator, reads address generator, write address generator, compression module and decompression module;
Described analysis module receives image sequence and analyzes, and the view data after analyzing is sent to described inverse discrete cosine transform module and described motion compensated predictor;
Described inverse discrete cosine transform module carries out sending to described colligator behind the inverse discrete cosine transform to the view data of receiving;
Described motion compensated predictor is judged the type of coding of this two field picture in the view data that receives;
When this two field picture is the independent image frame, described motion compensated predictor indicates described write address generator to generate a write address, the view data of the direct self-converse discrete cosine transform module in the future of described colligator is used for output as this frame image data and shows, this frame image data that described in addition colligator also will be used to export sends described compression module to as the reference two field picture, described compression module compresses described reference frame image, and the reference frame image after the compression is written into described memory module according to described write address;
When this two field picture is forward predicted frame, the indication of described motion compensated predictor is described reads address generator and generates one and read the address, also indicate described write address generator to generate a write address, read needs are read in the address from described memory module forward reference frame image according to described, described decompression module sends described colligator to after the forward reference frame image that reads is decompressed, described colligator in the future the view data of self-converse discrete cosine transform module combine with the forward reference frame image of described decompression and will in conjunction with after data be used for exporting demonstration as this frame image data, this frame image data that described in addition colligator also will be used to export sends described compression module to as the reference two field picture, described compression module compresses described reference frame image, and the reference frame image after the compression is written into described memory module according to described write address;
When this two field picture is two-way reference frame, the indication of described motion compensated predictor is described reads address generator and generates one and read the address, from described memory module, read described forward and backward reference frame image according to the described address of reading, described decompression module sends described colligator to after the forward and backward reference frame image that reads is decompressed, described colligator in the future the view data of self-converse discrete cosine transform module combine with the forward and backward reference frame image of described decompression and will in conjunction with after data be used for exporting demonstration as this frame image data.
2. motion compensating system as claimed in claim 1 is characterized in that described motion compensating module also comprises read command register, write command register;
Described motion compensated predictor is sent indication by described read command register to reading address generator;
Described motion compensated predictor is sent indication by described write command register to the write address generator.
3. motion compensating system as claimed in claim 2 is characterized in that described motion compensating module also comprises adjuster, half pixel digital difference detector, read buffer register;
Reference frame image data after decompression module decompresses through being sent in the described read buffer register after described adjuster, the half pixel digital difference detector, are read the reference frame image data of described decompress(ion) successively from described read buffer register by described colligator.
4. motion compensating system as claimed in claim 1 is characterized in that comprising in the described compression module device that this frame image data after the described combination is carried out discrete cosine transform (DCT);
The device that ac coefficient in the discrete cosine transform coefficient is expanded;
To the device that quantizes through the resulting coefficient of described discrete cosine transform; And
The device that discrete cosine transform coefficient after quantizing is compressed;
Described decompression module comprises the device that described reference frame image data are decompressed;
Reference frame image data after decompressing are carried out the device of de-quantization;
The ac coefficient of de-quantization is carried out the device of contrary expansion; And
To carried out the device of inverse discrete cosine transform (IDCT) by the reference frame image data of de-quantization.
5. method of carrying out motion compensation, this method comprises:
Image sequence is carried out inverse discrete cosine transform;
Type of coding to this two field picture in the image sequence is judged;
When this two field picture is the independent image frame, indication generates a write address, directly the view data behind the inverse discrete cosine transform being used for output as this frame image data shows, this frame image data that will be used in addition export compresses as the reference two field picture, and the reference frame image after the compression is written into memory module according to described write address;
When this two field picture is forward predicted frame, indication generates one and reads address and a write address, from described memory module, read the forward reference frame image and decompress according to the described address of reading, view data behind the inverse discrete cosine transform combined with the forward reference frame image of described decompression and will in conjunction with after data be used for output as this frame image data and show, this frame image data that will be used in addition export compresses as the reference two field picture, and the reference frame image after the compression is written into described memory module according to described write address;
When this two field picture is two-way reference frame, indication generates one and reads the address, from described memory module, read the forward and backward reference frame image and decompress according to the described address of reading, the view data behind the inverse discrete cosine transform is combined with the forward and backward reference frame image of described decompression and will in conjunction with after data be used for output as this frame image data and show.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200510123882 CN100481954C (en) | 2005-11-23 | 2005-11-23 | System and method for motion compensating |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200510123882 CN100481954C (en) | 2005-11-23 | 2005-11-23 | System and method for motion compensating |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1787643A CN1787643A (en) | 2006-06-14 |
| CN100481954C true CN100481954C (en) | 2009-04-22 |
Family
ID=36784906
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200510123882 Active CN100481954C (en) | 2005-11-23 | 2005-11-23 | System and method for motion compensating |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100481954C (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101616321B (en) * | 2008-06-26 | 2014-09-24 | 晨星软件研发(深圳)有限公司 | Circuit and method for processing image |
-
2005
- 2005-11-23 CN CN 200510123882 patent/CN100481954C/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN1787643A (en) | 2006-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6421385B1 (en) | Apparatus and method for efficient conversion of DV (digital video) format encoded video data into MPEG format encoded video data by utilizing motion flag information contained in the DV data | |
| CN101690163B (en) | Shutter time compensation | |
| JP3365771B2 (en) | Video signal compression device | |
| CN102792689B (en) | Delta compression can be carried out and for by image, remote display is presented to the amendment of estimation and metadata | |
| JP3462208B2 (en) | Video signal decompressor for independently compressed even and odd field data | |
| KR100781629B1 (en) | A method for reducing the memory required for decompression by storing compressed information using DCT base technology and a decoder for implementing the method | |
| CN101889447B (en) | Extension of the AVC standard to encode high resolution digital still pictures in series with video | |
| US20090141809A1 (en) | Extension to the AVC standard to support the encoding and storage of high resolution digital still pictures in parallel with video | |
| US20030095603A1 (en) | Reduced-complexity video decoding using larger pixel-grid motion compensation | |
| KR100192696B1 (en) | Method and apparatus for reproducing picture data | |
| US7113644B2 (en) | Image coding apparatus and image coding method | |
| EP1940175A1 (en) | Image encoding apparatus and memory access method | |
| KR100312421B1 (en) | A conversion method of the compressed moving video on the video communication system | |
| KR100681242B1 (en) | Method of decoding moving picture, apparatus for decoding moving picture and system-on-a-chip having the same | |
| CN101888545B (en) | Compression coding method for signal source with low code rate | |
| JP3850015B2 (en) | Digital video signal recording / reproducing apparatus and transmission apparatus | |
| CN100481954C (en) | System and method for motion compensating | |
| JP2003230148A (en) | Image data coding unit | |
| JPH11136686A (en) | Decoded image conversion circuit and decoded image converter | |
| US20110249719A1 (en) | Video compression | |
| JP2003179826A (en) | Image reproducing and displaying device | |
| JP3141149B2 (en) | Image coding device | |
| JP2009071802A (en) | Dynamic image encoding method and device, and imaging system | |
| JP2005159832A (en) | Signal processor | |
| CN101917616A (en) | The method and system that is used for digital coding three-dimensional video image |
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 | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20180409 Address after: 100191 Xueyuan Road, Haidian District, Haidian District, Beijing, No. 607, No. six Patentee after: Beijing Vimicro AI Chip Technology Co Ltd Address before: 100083, Haidian District, Xueyuan Road, Beijing No. 35, Nanjing Ning building, 15 Floor Patentee before: Beijing Vimicro Corporation |