CN2796278Y - DVD decoder - Google Patents

Abstract

The utility model relates to an H264 decoder for DVDs, which comprises an inner-frame frequency detection selecting unit, an inner-frame predicting unit, a movement displacement estimating unit, a movement compensating unit, an inverse-quantification unit, an inverse-conversion unit, and a frame reconstructing unit, wherein the inner-frame frequency detection selecting unit divides a current frame Fn into macro blocks which have an inner-frame mode and an inter-frame mode, the inner-frame predicting unit uses the encoded macro blocks in the current frame for prediction for the inner-frame mode, the movement displacement estimating unit and the movement compensating unit use one or a plurality of frames as the reference for the inter-frame mode, and estimate the ME and the movement compensation MC by movement displacement for movement prediction, the inverse-quantification unit and the anti-conversion unit carry out the inverse quantization Q<-1 > and the inverse transformation T <-1 > for a quantization coefficient X, and the frame reconstructing unit adds predicting coefficients to obtain a uF frame which is not filtered. The uF frame is filtered to obtain a current reconstructing frame Fn'. The decoder can be suitable for the requirements of high image quality and the code rate application of low speed.

Description

The DVD decoder

Technical field

The utility model relates to the video compression decode technology, refers to a kind of video compression decode method and device of the DVD of being used for high definition decoding deck especially.

Background technology

Digital video technology has obtained increasingly extensive application in communication and audiovisual field, particularly since 90 years 20th century, along with the fast development of Internet and mobile communication, video information and multimedia messages processing and the transmission technology in Internet and mobile network becomes focus in the informationization of current China.

As everyone knows, video information has series of advantages, as intuitive, certainty, high efficiency, popularity etc.But the amount of information of coded video information is too big, and video is effectively used, and must at first solve the video compression coding problem, secondly solves compression rear video problem for quality assurance.

For this reason, people have paid huge hard work, have nowadays born great successes.Since announcing the 1st video coding international standard from CCITT in 1984, in up to now 21 years, International Standards Organization such as ITU-T have issued 10 video coding international standards nearly successively, promoted the development of video communication and audio visual technology greatly, but say that strictly this two big field development so far of video communication and audiovisual can not be satisfactory, stack up is said, the scope of using is wideless, mainly is that the contradiction between video compression and the quality can not well solve.For example: video telephone is considered to a kind of desirable communication equipment always, can fail so far to popularize over nearly 30 years, is exactly because the ratio of performance to price is not high.

1984, CCITT the 15th seminar issued digroup video conference coding standard and has H.120 advised.H.261, CCITT in 1988 has passed through the video coding standard to advise, is called as a milestone of video compression coding.From then on, ITU-T, the mixed coding method during H.261 the coding method based on a series of video encoding standards of waveform of announcements such as ISO all is based on.

1986, ISO and CCITT set up the international expert group of associating.The static international compression algorithm international standard of research continuous tone has been passed through Joint Photographic Experts Group in July, 1992.

1988, ISO/IEC information technology joint committee set up motion picture expert group version.Announced the MPEG-1 video encoding standard in 1991, code check is 1.5Mbit/S, is mainly used in the video compression of the VCD of family.In November, 1994, announced Moving Picture Experts Group-2, be used for digital video broadcasting (DVB), the video compression of family expenses DVD and high definition TV (HDVT).Code check, is respectively applied in the video compression of various criterion and different stage until 100Mbit/S from 4Mbit/S, 15Mbit/S.

Nineteen ninety-five, ITU-T releases H.263 standard, is used to be lower than the low bit-rate video transmission of 64Kbit/S, as the video conferencing in the PSTN channel, multi-media communication etc.2000, announced respectively H.263+ again, H.263++ wait standard.

In December, 1999, ISO/IEC has passed through that " coding standard of audiovisual object---MPEG-4, it has also emphasized the interactivity and the flexibility of multimedia communication except definition audiovisual compressed encoding standard.

In March, 2003, ITU-T/ISO formally announces H.264 video compression standard, because it has than the outstanding performance of standard in the past, is called video encoding standard of new generation by people.Particularly, with H.263 or MPEG-4 compare, under equal quality, its numeric code rate can reduce about half, under same code check, its letter price ratio can be improved significantly in other words.Thus, H.264 standard is subjected to paying attention to widely and promoting in the world, it has not only significantly improved compression ratio, and has a good network compatibility, be widely used in network service, multimedia communication, the HDTV (High-Definition Television), the performance that equally also is high definition DVD improves with universal, has brought revolutionary opportunities and challenges.Yet, as high-end audio-visual products---DVD, how to adopt H.264 coding standard, become a problem demanding prompt solution.

The utility model content

In view of this, the purpose of this utility model is to provide the H.264 decoder of a kind of DVD of being applicable to, and this decoder can adapt to more high image quality and low speed code check demands of applications.

For achieving the above object, a kind of DVD decoder is provided, it is characterized in that comprising:

Frequently survey selected cell in the frame, Fn is divided into macro block present frame, and macro block has in the frame and two kinds of patterns of interframe;

Intraprediction unit, itself and the interior frequency of frame are surveyed selected cell and are linked in sequence, and use the macro block of having encoded in the present frame that the frame mode of frequently surveying in the selected cell in the frame is predicted;

Interconnective moving displacement estimation unit and motion compensation units, it also with in the frame is frequently surveyed selected cell and is connected, and uses one or more in the past frames as a reference, by moving displacement estimation ME and motion compensation MC, to frequently surveying the inter-frame mode in the selected cell in the frame, carry out motion prediction;

Converter unit and quantifying unit link to each other with motion compensation units with the moving displacement estimation unit, and the difference of its predicted value and original value is carried out conversion T, quantized Q;

Rearrangement and entropy coding unit link to each other with quantifying unit with converter unit, and the signal after its quantification is resequenced and entropy coding;

Inverse quantization unit and inverse transformation block also link to each other with quantifying unit with converter unit, and its quantization parameter X is carried out re-quantization Q ^-1, inverse transformation T ^-1

The reconstructed frame unit links to each other with inverse transformation block with inverse quantization unit, and links to each other with motion compensation units with the moving displacement estimation unit, and itself and predictive coefficient addition obtain the uF ' frame of non-filtered, and uF ' frame is carried out interblock filtering, obtain current reconstructed frame Fn '.

According to preferred embodiment of the present utility model, wherein said H.264 decoder also comprises filter, and resulting uF ' frame is carried out filtering.

Description of drawings

In order to make the purpose of this utility model, technical scheme and advantage clearer, below in conjunction with accompanying drawing the utility model is described in further detail, wherein:

Fig. 1 is the H.264 schematic diagram of the high definition DVD decoding deck of coding/decoding system of employing of the present utility model; Fig. 2 is the block diagram of decoding master chip AM3418 of the present utility model;

Fig. 3 is the schematic diagram of H.264 coding/decoding system of the present utility model;

Fig. 4 is the realization flow figure of the utility model video compression decode method;

Fig. 5 is the schematic diagram of lamella code flow structure of the present utility model;

Fig. 6 is the schematic diagram of the frame storage configuration of the utility model and prior art;

Fig. 7 is the schematic diagram of global variable structure of the present utility model;

Fig. 8 is the schematic diagram of INTRA16 of the present utility model * 16 macro blocks decoding;

Fig. 9 is the schematic diagram of INTRA4 of the present utility model * 4 macro blocks decoding;

Figure 10 is the flow chart of INTRA macro block decoding of the present utility model;

Figure 11 is 1/4 picture element interpolation of the present utility model;

Figure 12 is the more detailed block diagram of H.264 encoder of the present utility model; With

Figure 13 is the more detailed block diagram of H.264 decoder of the present utility model.

Embodiment

As everyone knows, the emphasis of MPEG-4 technology is flexibly, has favorable expansibility, can carry out the expansion in time domain and spatial domain.MPEG-4 is in order to support high-new compression, content-based mutual and content-based grading extension.And H.264 support the application of 3 different class:

1., basic class: be mainly used in " video session "

2., expansion class: the Video show that is mainly used in network

3., main class: be mainly used in consumer-electronics applications, as digital television broadcasting, digital video is stored, is play---the DVD product.

H.264/AVC standard has become Digital Television, video communication since it is published.The network media, the leading technology in fields such as multi-media SMS breath, and all successfully importings in above field have been created tangible economic benefit, and its electronic product that is applied to and transmission quality are greatly improved.Owing to H.264 and ambiguously stipulate how an encoder is realized, but the sentence structure of a video bit stream of having compiled sign indicating number and the coding/decoding method of this bit stream have been stipulated.The encoder of each manufacturer should be able to intercommunication under this framework, has bigger flexibility in realization, therefore is all the DVD technology in audiovisual field, possesses an importing tangent condition H.264 fully, and technical is very reliable.Yet how to import H.264, need on the coding and decoding process of DVD, from hardware to software, do very careful design and improvement.

Referring to Fig. 1, high definition DVD decoding deck of the present utility model adopts H.264 coding/decoding system.This DVD decoding deck comprises at least: electric-motor drive unit, and servo control unit VT7208, decoding and DSP handle AM3418, video processing unit, audio treatment unit, SDRAM4X16M, FLASHROM, and EPROM29LV800.The wherein feeding of the motor driven systems control shaven head of master chip AM3418 control and the rotation of main shaft, SERVO CONTROL chip VT7208 finishes the focusing and the tracking control of shaven head, detection and automatic gain system are also finished by VT7208 simultaneously, thereby obtain the correct RF signal of signal amplitude and phase place, after VT7208 handles demodulation, obtain audio-video signal again.Signal after the demodulation obtains 64 digital signal via the DSP signal processing.Signal is stored by SDRAM, and buffering is also carried out encoding and decoding via warp let-off AM3418 behind the digital filtering and handled.In the present embodiment, all runnings of system and control all are that the microcode of being stored by EPROM29LV800 is carried out.

With reference to Fig. 2, via the inner D/A conversion of AM3418, video is exported via decoded digital audio-video signal, video component output, audio frequency output, digital audio and video signals and the output of various required analogue component.

With reference to Fig. 3, show H.264 coding/decoding system of the present utility model, wherein all coding/decoding systems are all finished by master chip AM3418.A video coding and decoding system comprises decoding, the playback of collection, coding transmission and the image of image.In the H.264 coding/decoding system of Fig. 3, the RF signal inputs to the digital filtering network via demodulation and Digital Signal Processing (DSP) afterwards.The digital filtering network carries out digital filtering to this signal, to obtain suitable digital signal input video encoder.In encoder H.264, finish the infra-frame prediction of input signal, estimation and exented memory control.

H.264/AVC introduced some new compression methods in, to improve compression efficiency, it has following characteristics:

(1) the motion compensation performance of Zeng Qianging.Adopt the motion compensation of tree and the motion vector prediction technology of 1/4 pixel precision, make predictive frame, reduce mutual difference, remove temporal data redundancy, improve the compression ratio of coding more near primitive frame;

(2) adopting more, fritter carries out transition coding.8 * 8DCT conversion meeting in the past causes the coefficient distortion, makes between reconstructed frame and primitive frame to have error.H.264 use 4 * 4 integer transforms, guarantee that the inverse transformation coefficient is undistorted, and reduced computational complexity;

(3) adopt the interblock filter to improve performance.Because coding diminishes, can cause that the later interblock jump in brightness of reconstruct is big, phenomenons such as mosaic appear in image, influence people's visual experience.Use filter at interblock,, can make the image after the reconstruct more press close to original image with the jump in brightness between smooth block;

(4) H.264 high performance entropy coding uses the elongated degree entropy coding of content-adaptive because the DCT coefficient correlation of interblock is big, has reduced the correlation redundancy of interblock, improves code efficiency;

(5) adopt SP and SI frame to support to switch between video flowing.H.264 introduce the notion of switch frame SP and SI.Decoder can use switch frame according to the situation of current network, effectively switches between the video flowing of different quality, solves predictive frame in the handoff procedure for want of reference frame and the decoding error that causes.

Hereinafter with reference to Figure 12, Figure 13, to H.264 encoder of the present utility model and H.264 decoder be described in more detail.

As shown in figure 12, during coding, at first present frame Fn is divided into macro block.Macro block has in the frame and two kinds of patterns of interframe.Frame mode uses the macro block of having encoded in the present frame to predict; Inter-frame mode uses one or more in the past frames as a reference, estimates ME and motion compensation MC by moving displacement, carries out motion prediction.Then, the difference of predicted value and original value is carried out conversion T, quantized Q, rearrangement and entropy coding.Quantization parameter X is carried out re-quantization Q ^-1, inverse transformation T ^-1After, with the predictive coefficient addition, obtain the uF ' frame of non-filtered, uF ' frame is carried out interblock filtering, obtain current reconstructed frame Fn '.

As shown in figure 13, decode procedure is relatively simple, carries out contrary operation, promptly exportable reconstructed image data corresponding to the each several part of encoder.

Hereinafter with reference to Fig. 4 to Figure 11, do detailed elaboration from the optimization of framework, data structure and flow process thereof, code table and the program of decoder H.264 respectively.

With reference to Fig. 4, H.264 the decoder overall framework mainly comprises each module of primary module, inlet module and Slice, macro block decoding.Decoding can be divided into 3 main standalone modules usually: code fetch, decoding, output (comprising reprocessing).

The code fetch module comprises from code stream reads in buffering with data, and obtains current code value from buffering, in whole flow process, to recycle, and be a key modules that influences decoding speed.Can before a frame decoding, read in the code stream of enough frames for decoding; Also can set a certain size buffering, and when each code fetch, judge whether to arrive buffering ending place, as then upgrading buffer content to ending place.Before a kind of scheme require buffering bigger, the also difficult size of estimating a frame code stream, but if Installed System Memory is enough, this kind scheme is adopted in suggestion, can avoid in second kind of scheme judgement during each code fetch, thereby influence decoding speed.

Decoder module is the core in the entire decoder, and in encoding code stream H.264, Slice is the unit of minimum byte-aligned, and the coding structure of a Slice as shown in Figure 5.In the Slice header structure, comprise frame number, image size, reference frame number, quantization parameter information such as (QP), comprise macro block (mb) type, intra prediction mode, motion vector (MVD) in the macro block head, CBP (Coded Block Pattern), quantization parameter fine setting information such as (delta-QP).Determined the flow process of whole decoding to be thus: to read Slice header → read macro block head (finishing infra-frame prediction or motion compensation) → residual coefficients decoding.Decode procedure as shown in Figure 4.

Image output module comprises conversion, image interpolation of image filtering, image output format etc., and this is not the unique portion of decoder H.264, does not repeat them here.

Fig. 6 shows the storage configuration of present frame and reference frame.Wherein Fig. 6 a is the frame storage configuration of existing jm50c, and Fig. 6 b is the frame storage configuration of the utility model decoder, to the frame that the brightness and the colourity of every frame adds a certain size respectively, the design of can simplifying procedures.

In motion compensation, 1/2 picture element interpolation is obtained by (1 ,-5,20,20 ,-5,1) weighted median filtering by 6 pixel brightness values of periphery.Borderline picture element interpolation around considering needs only the most close borderline pixel value of copy to the frame among Fig. 6 b.Like this, having simplified programming to a certain extent, is that program structure is clear.To the image edged, also can support the unrestricted motion compensation.

Same structure is adopted in depositing of reference frame and present frame, after a frame decoding is finished, as long as the initial storage address of exchange reference frame and present frame, can avoid the mobile of data of determining greatly as jm50c like this, has accelerated the speed of decoding to a certain extent.Its program is as follows.

Tmp_frame＝frame_ref；

frame_ref＝frame_curr；

frame_curr＝tmp_frame；

Program is deposited in all global variables in three structures, as shown in Figure 7.Wherein H264 TABLES comprises the used code tables of all decodings, deposit with array, and before the decoding beginning income internal memory, be equivalent among the jm50c code table be defined in each program module respectively, decoding efficiency increases.

H264 IMAGE comprises SLICE_HEADER and two structures of MAC-ROBLOCK, and wherein SLICE_HEADER deposits the used global variable of the current Slice of decoding, the normally global variable that obtains from the Slice head.MACROBLOCK is the global variable of current macro normally then.At main program all global variables are distributed the monoblock internal memory, help to improve the internal memory service efficiency.

It should be noted that above-mentioned global variable data are for the convenience on programming, as considering multipath decoding etc., be the conflict of avoiding naming, except that TABLES structure H.264, other data suggest adopts function parameters transmission method to realize, does not adopt global variable.

In jm50c, I, P, same module is adopted in the decoding of B frame, and this needs a large amount of I, P, B frame and INTRA thereof, the repeating of INTER macro block judged, makes decoding speed limited.For this reason, and consider extensibility, decoder of the present utility model is with the independent respectively decoding of I frame, P frame (wouldn't comprise the B frame).The macro block decoding also is divided into INTRA16*16, INTRA4*4, three decoder modules of IN-TER, judge that with repeating in the middle of saving improved decoding speed effectively, it is more clear that program structure also seems, and, help the expansion of program mould because each module ground is relatively independent.Below introduce all kinds of macro-block decoding process respectively:

Because infra-frame prediction is the direct prediction of pixel intensity or chromatic value, it and coefficient inverse transformation are two processes fully independently, therefore, for the INTRA16*16 macro block, as shown in Figure 8, at first finish infra-frame prediction, the conversion coefficient of decoding then, intra prediction value is directly added memory image ground, ground noted earlier internal memory, at last the value after the integral inverse transform is added, promptly finish a macro block ground decoding; For the INTRA4*4 macro block, as shown in Figure 9, be the decode procedure of the piece of a 4*4, based on the same INTRA16*16 of method,, finish the decoding of a macro block by 16 circulations.

For the INTER macro block, as shown in figure 10, mainly comprise the conversion coefficient decoding of prediction, macro block reconstruction and the motion residual of motion vector.Motion vector prediction comprises macro block mode and sub-macro block mode, and macro block is rebuild 1/4 picture element interpolation, and the prediction mode complexity directly influences decoding speed, and the optimization on each details of program seems particularly important.

For 1/4 picture element interpolation in the macro block reconstruction, as shown in figure 11, G, H, M, N is whole pixel, other a, b ..., r is respectively 1/2 or 1/4 picture element interpolation, a, b, ..., its interpolation method of r has nothing in common with each other, and present embodiment has adopted function pointer, adopt independently function to finish respectively to each interpolation, increased the size of code to a certain extent, but effectively improved decoding speed.

H.264 coding standard is with the different of video compression standard maximum of adopting in the past, has adopted multi-reference frame to predict in motion estimation process and has improved precision of prediction.Multi-reference frame prediction is built the buffer memory of a store M reconstruction frames at the encoding and decoding end with exactlying, current block to be encoded can all reconstruction frames in buffer memory in the match block of searching optimum carry out motion compensation.

H.264 coding and decoding technology of the present utility model is applicable to the video compression of DVD, and has following technical characterstic: support interlaced video, adopt the interframe encode of B sheet and the intraframe coding of employing weight estimation; Support to utilize based on contextual adaptive arithmetic coding (CABAC); Support to utilize based on contextual adaptive Variable Length Code (CAVLC); Obtain high video compression ratio, be about 2 times of MPEG-4; Based on the motion compensated predictive coding of inter prediction, can reconstruct identical image with reference to the different parameters frame---be SP frame coding; Adopt the adaptive boundary filter, some level filter, the chip level filter is realized deblocking filtering.

The key technical indexes of the present utility model is: reference frame can reach 5～15 frames, saves the bit rate of 5%-10%; The fast algorithm of adaptable search scope is 287 times of global search algorithm, and three go on foot 2.8 times of searching algorithms; Use 1/4 mark as the rope precision search; Channel bit error rate≤10-4; Support the Code And Decode of the continuous or interlaced video of 4:2:0; The precision of motion vector 〉=1/4 pixel; Brightness maximum QP=51, colourity maximum QP=39; This improves 3dB than MPEG-4 the noise of output YC component.

By parameters contrast, environmental test, the DVD product after importing H.264, cost does not have too many cost to be increased, but performance and video quality be greatly improved, and can mate with Digital Television fully, forms real digital home entertainment.H.264 importing can make the really progressive high definition epoch of DVD product.

Although by reference some preferred embodiment of the present utility model, the utility model is illustrated and describes, but those of ordinary skill in the art is to be understood that, can make various changes to it in the form and details, and the spirit and scope of the present utility model that do not depart from appended claims and limited.

Claims (2)

1, a kind of DVD decoder is characterized in that comprising:

Frequently survey selected cell in the frame, Fn is divided into macro block present frame, and macro block has in the frame and two kinds of patterns of interframe;

Intraprediction unit, itself and the interior frequency of frame are surveyed selected cell and are linked in sequence, and use the macro block of having encoded in the present frame that the frame mode of frequently surveying in the selected cell in the frame is predicted;

Interconnective moving displacement estimation unit and motion compensation units, it also with in the frame is frequently surveyed selected cell and is connected, and uses one or more in the past frames as a reference, by moving displacement estimation ME and motion compensation MC, to frequently surveying the inter-frame mode in the selected cell in the frame, carry out motion prediction;

Converter unit and quantifying unit link to each other with motion compensation units with the moving displacement estimation unit, and the difference of its predicted value and original value is carried out conversion T, quantized Q;

Rearrangement and entropy coding unit link to each other with quantifying unit with converter unit, and the signal after its quantification is resequenced and entropy coding;

Inverse quantization unit and inverse transformation block also link to each other with quantifying unit with converter unit, and its quantization parameter X is carried out re-quantization Q ^-1, inverse transformation T ^-1

The reconstructed frame unit links to each other with inverse transformation block with inverse quantization unit, and links to each other with motion compensation units with the moving displacement estimation unit, and itself and predictive coefficient addition obtain the uF ' frame of non-filtered, and uF ' frame is carried out interblock filtering, obtain current reconstructed frame Fn '.

2. decoder according to claim 1 also comprises filter, and resulting uF ' frame is carried out filtering.

Images