CN101588498A - Video image data compression and decompression method and device - Google Patents

Abstract

The invention provides a video image data compression method, comprising the following steps: partitioning for the video image data; adopting a compression algorithm to compress the present block of the video image data to obtain a code stream block; adopting a decoding algorithm corresponding to the compression algorithm to decode the code stream block to obtain the rebuilding data of the present block; calculating the distortion degree of the rebuilding data relative to the present block before compressing and storing the distortion flag bit reflecting the distortion degree to the code stream block. The invention solves the problem in the prior art that when the former frame image data information is adopted to execute the video processing, for example, in the motion estimation and movement compensation, the former frame data information usually obtain the data blocks with different rebuilding quality only from the former frame fixed code stream block, which has larger influence on the moving estimate accuracy, even the error can appear.

Description

Video image data compression, decompression method and device

Technical field

The present invention relates to image processing field, in particular to a kind of video image data compression, decompression method and device.

Background technology

In video image data compression,, adopt lossy compression method usually in order to obtain higher compression ratio; In order to reduce computational complexity or hardware spending, need carry out piecemeal to image usually, the piecemeal as 8 * 8,32 * 32 piecemeal etc.; Compression ratio in order to obtain to fix often requires each piecemeal to have identical compression ratio simultaneously, and owing to contained high and low frequency component difference in the different piecemeals in the image, the complexity of its compression is also different.According to rate distortion theory, when requiring distortion to fix, then compression ratio is unfixing; Otherwise when requiring compression ratio to fix, then distortion is unfixing.

In realizing process of the present invention, the inventor finds in the prior art when needs are used the former frame image data information and carried out Video processing, in estimation and motion compensation, the data message of former frame usually can only be from the fixing code stream block of former frame obtains the data block that reconstruction quality is not quite similar (such as when taking the low frequency compression algorithm through decompress(ion), can obtain higher decompression reconstruction quality for low frequency background piece, the high frequency texture block can only obtain lower decompression reconstruction quality), and this estimated accuracy to motion has bigger influence, even mistake can occur.

Summary of the invention

The present invention aims to provide a kind of video image data compression, decompression method and device, can solve in the prior art when needs are used the former frame image data information and carried out Video processing, in estimation and motion compensation, the data message of former frame can only obtain the data block that reconstruction quality is not quite similar through decompress(ion) usually from the code stream block that former frame is fixed, and this estimated accuracy to motion has bigger influence, even the problem of mistake can occur.

In an embodiment of the present invention, provide a kind of video image data compression method, may further comprise the steps:

Vedio data is carried out piecemeal;

The employing compression algorithm is compressed the current block of vedio data, obtains code stream block;

Adopt the decoding algorithm corresponding that code stream block is decoded, obtain the data reconstruction of current block with compression algorithm;

Calculate data reconstruction with respect to the compression of current block before the distortion value of data, and will reflect that the distortion flag bit of above-mentioned distortion value stores in the code stream block.

Preferably, in above-mentioned video image data compression method, compression algorithm is following any compression algorithm: BTC algorithm, JPEG Coding Compression Algorithm, and based on one dimension or two-dimentional piecemeal compression algorithm, scalar quantization algorithm, Vector Quantization algorithm and the grid coding quantization algorithm of wavelet transformation.

Preferably, in above-mentioned video image data compression method, calculate data reconstruction with respect to the compression of current block before the distortion level of data specifically comprise:

Calculate the error or the coefficient correlation of the preceding data of compression of data reconstruction and current block;

The threshold value of error or coefficient correlation and setting is compared, obtain the distortion level of the packed data of current block, and represent above-mentioned distortion level with limited bit.

Preferably, in above-mentioned video image data compression method, error is worst error or mean error or mean square error.

In an embodiment of the present invention, also provide a kind of vedio data decompression method, may further comprise the steps:

From code stream block, obtain the distortion flag bit of reflection code stream block compression losses;

If the distortion flag bit is for poor, when estimation, utilize the motion vector interpolation estimation of the adjacent block of current block to obtain the motion vector of current block.

Preferably, in above-mentioned vedio data decompression method, further comprising the steps of:

If the distortion flag bit is for well, utilize code stream block to recover the vedio data of current block, when estimation, the current frame data that utilizes current block and the corresponding decompression block data with better quality estimate to obtain the motion vector of current block by block matching motion.

In an embodiment of the present invention, also provide a kind of video image data compression device, having comprised:

Coding module is used for vedio data is carried out piecemeal, and the current block of vedio data is compressed, and obtains code stream block;

Decoder module is used for code stream block is decoded, and obtains the data reconstruction of current block;

Evaluation module, be used to calculate data reconstruction with respect to the compression of current block before the distortion level of data, and will reflect that the distortion flag bit of distortion level stores in the code stream block.

In an embodiment of the present invention, also provide a kind of vedio data decompressing device, having comprised:

Parsing module is used for obtaining the distortion flag bit that reflects the code stream block compression losses from code stream block;

Rebuilding module if be used for the distortion flag bit for poor, when estimation, utilizes the motion vector interpolation estimation of the adjacent block of current block to obtain the motion vector of current block.

Preferably, in above-mentioned vedio data decompressing device, also comprise:

Decompressing device, be used for when the distortion flag bit has been, utilize described code stream block to recover the vedio data of described current block, when estimation, the current frame data that utilizes current block and the corresponding decompression block data with better quality estimate to obtain the motion vector of current block by block matching motion.

In the above-described embodiments, by after vedio data is compressed, promptly the compressed code flow piece is decoded, calculate the distortion level of decoded data reconstruction with respect to the preceding data of compression, and the distortion flag bit that will reflect distortion level is stored in the code stream block, make when decompressor end is carried out the vedio data processing, can make judgement to the compression quality of code stream block according to the distortion value that is stored in the code stream block, thereby adjust follow-up video processnig algorithms for example in estimation and motion compensation, when the compression quality flag bit of current block is difference, can estimate to obtain by interpolation according to the motion vector of its contiguous block, thereby can obtain current block motion vector comparatively accurately, overcome in the prior art needs use the former frame image data information carry out Video processing the time, in estimation and motion compensation, the data message of former frame can only obtain the data block that reconstruction quality is not quite similar through decompress(ion) usually from the code stream block that former frame is fixed, and this estimated accuracy to motion has bigger influence, even the problem of mistake can occur.

Description of drawings

Accompanying drawing described herein is used to provide further understanding of the present invention, constitutes the application's a part, and illustrative examples of the present invention and explanation thereof are used to explain the present invention, do not constitute improper qualification of the present invention.In the accompanying drawings:

Fig. 1 shows video image data compression method flow chart according to an embodiment of the invention;

Fig. 2 shows vedio data decompression method flow chart according to an embodiment of the invention

Fig. 3 shows video image data compression apparatus module figure according to an embodiment of the invention;

Fig. 4 shows vedio data decompressing device module map according to an embodiment of the invention;

Two frame data utilization block matching algorithm carried out the hardware schematic diagram of estimation of motion vectors before and after Fig. 5 showed and utilizes according to an embodiment of the invention.

Embodiment

Below with reference to the accompanying drawings and in conjunction with the embodiments, describe the present invention in detail.

Fig. 1 shows video image data compression method flow chart according to an embodiment of the invention, may further comprise the steps:

S102 carries out piecemeal to vedio data;

S104, the employing compression algorithm is compressed the current block of vedio data, obtains code stream block;

S106 adopts the decoding algorithm corresponding with compression algorithm that code stream block is decoded, and obtains the data reconstruction of current block;

S108, calculate data reconstruction with respect to the compression of current block before the distortion level of data, and will reflect that the distortion flag bit of distortion level stores in the code stream block.

In the present embodiment, by after vedio data is compressed, immediately the compressed code flow piece is decoded, calculate the distortion level of decoded data reconstruction with respect to the preceding data of compression, and the distortion flag bit that will reflect distortion level is stored in the code stream block, make when decompressor end is carried out the vedio data processing, can adjust follow-up video processnig algorithms according to the distortion flag bit that is stored in the code stream block, for example in estimation and motion compensation, when the compression quality flag bit of current block is difference, can estimate to obtain by interpolation according to the motion vector of its contiguous block, thereby can obtain current block motion vector comparatively accurately, overcome in the prior art when estimation and motion compensation, the data message of former frame can only obtain the data block that reconstruction quality is not quite similar through decompress(ion) usually from the code stream block that former frame is fixed, and the estimated accuracy of motion is had bigger influence, even Problem-Error can appear.

Preferably, in above-mentioned video image data compression method, compression algorithm is BTC algorithm, JPEG Coding Compression Algorithm, based on one dimension or two-dimentional piecemeal compression algorithm, scalar quantization algorithm, Vector Quantization algorithm and the grid coding quantization algorithm of wavelet transformation.

Preferably, in above-mentioned video image data compression method, calculate data reconstruction with respect to the compression of current block before the distortion value of data specifically comprise:

Calculate the error or the coefficient correlation of the preceding data of compression of data reconstruction and current block;

The threshold value of error or coefficient correlation and setting is compared, obtain the distortion value of the packed data of current block.

Preferably, in above-mentioned video image data compression method, error is worst error or mean error or mean square error.

The formula of worst error is: MaxBias=Max{|X _i-X _i' |;

The formula of mean error is: $\mathrm{MeanBias}=\frac{\underset{i=1}{\overset{8}{\mathrm{\Σ}}}|X_i-{X_i}^{\′}|}{8};$

The formula of mean square error is: $\mathrm{MSE}=\frac{\underset{i=1}{\overset{8}{\mathrm{\Σ}}}{|X_i-{X_i}^{\′}|}^2}{8};$

X wherein _iFor compressing preceding view data, X _i' for compressing the view data that the back rebuilds.

By judging that worst error MaxBias or mean error MeanBias or mean square error MSE whether greater than threshold value, come the quality of data compression in the decision block, if represent with 1bit, it is good with bad just can only to represent:

Adopt the compression quality assessment formula of worst error:

$\mathrm{Quality}=\left\{\begin{array}{ccc}{1}^{\&prime;}b1& \mathrm{if}& \mathrm{MaxBias}<\mathrm{Threshold}\_\max \mathrm{bias}\\ 1^{\&prime;}b0& \mathrm{if}& \mathrm{MaxBias}\&GreaterEqual;\mathrm{Threshold}\_\max \mathrm{bias}\end{array}\right.;$

Adopt the compression quality assessment formula of mean error:

$\mathrm{Quality}=\left\{\begin{array}{ccc}{1}^{\&prime;}b1& \mathrm{if}& \mathrm{MeanBias}<\mathrm{Threshold}\_\mathrm{meanbias}\\ 1^{\&prime;}b0& \mathrm{if}& \mathrm{MeanBias}\&GreaterEqual;\mathrm{Threshold}\_\mathrm{meanbias}\end{array}\right.;$

Adopt the compression quality assessment formula of mean square error:

$\mathrm{Quality}=\left\{\begin{array}{ccc}{1}^{\&prime;}b1& \mathrm{if}& \mathrm{MSE}<\mathrm{Threshold}\_\mathrm{mse}\\ 1^{\&prime;}b0& \mathrm{if}& \mathrm{MSE}\&GreaterEqual;\mathrm{Threshold}\_\mathrm{mse}\end{array}\right..$

Worst error MaxBias, mean error MeanBias, mean square error MSE can not reflect the compressed and decompressed extent of damage to the image block amount of information comprehensively.As to the grain details image, though loss is arranged on mean flow rate, estimate with MSE very poor, but visually be not see difference sometimes, MSE is very high sometimes on the contrary, but the grain details of image block visually but can be realized bigger deviation by fuzzy.So use MaxBias, it is not the comprehensive of ten minutes that MeanBias, MSE weigh compression artefacts.At this moment can adopt coefficient correlation γ to weigh: the coefficient correlation γ of compressed and decompressed image block data and original image blocks of data may be defined as:

$\mathrm{\&gamma;}=\frac{\underset{i=1}{\overset{8}{\mathrm{\&Sigma;}}}(X_i-\stackrel{\&OverBar;}{X})({X_i}^{\&prime;}-{\stackrel{\&OverBar;}{X}}^{\&prime;})}{\sqrt{\underset{i=1}{\overset{8}{\mathrm{\&Sigma;}}}{(X_i-\stackrel{\&OverBar;}{X})}^2\underset{i=1}{\overset{8}{\mathrm{\&Sigma;}}}{({X_i}^{\&prime;}-{\stackrel{\&OverBar;}{X}}^{\&prime;})}^2}}.$

If the decompressing image blocks of data is relevant more with the original image blocks of data, then coefficient correlation approaches 1 more.For lossless compress, then corresponding compressed and decompressed video data block is identical, and coefficient correlation is 1.0.

Weigh compression artefacts with coefficient correlation, the quality of compression performance just compares with coefficient correlation and threshold value, that is:

$\mathrm{Quality}=\left\{\begin{array}{ccc}{1}^{\&prime;}b1& \mathrm{if}& \mathrm{\&gamma;}<\mathrm{Threshold}\_\mathrm{corr}\\ 1^{\&prime;}b0& \mathrm{if}& \mathrm{\&gamma;}\&GreaterEqual;\mathrm{Threshold}\_\mathrm{corr}\end{array}\right..$

In general, MaxBias, MeanBias, what MSE reflected is the loss of low frequency component behind the compressed decompress(ion) of video data block, and coefficient correlation γ reflection is the loss of high fdrequency component behind the compressed decompress(ion) of video data block, can be in the application according to specific requirement or the MaxBias that selects the reflection low frequency to lose, MeanBias, MSE, or the coefficient correlation γ of reflection high frequency loss, perhaps the two is whole, and represents the compressed and decompressed distortion situation of current data block with 1-2bit, and adds in the code stream block.

Fig. 2 shows vedio data decompression method flow chart according to an embodiment of the invention, may further comprise the steps:

S202 obtains the distortion flag bit that reflects the code stream block compression losses from code stream block;

S204 is if the distortion flag bit for poor, when estimation, utilizes the motion vector interpolation estimation of the adjacent block of current block to obtain the motion vector of current block.

In the present embodiment, at first from code stream block, obtain reflecting the distortion flag bit of current block compression losses, and with of the integral body judgement of this flag bit as the current data block compression artefacts, and then decision is to the utilization of this piece decompressed data, if the distortion flag bit is for poor, when estimation, utilize the motion vector interpolation of the adjacent block of current block to estimate the motion vector of current block.

For example, at utilization block matching algorithm (Block Matching Algorithm, when BMA) carrying out the motion vector extraction, if the compression piecemeal piecemeal identical with the employing of block matching algorithm, and when the compression artefacts flag bit of former frame decompression block is difference, be incredible then, can simply come interpolation to estimate the motion vector of current block with near the reliable motion vector of piece according to the continuity of motion choosing motion vector with minimum SAD by calculating SAD (Sum of Absolute difference); Thereby the motion vector match that can avoid compression artefacts to make the mistake so just.According to the block matching algorithm principle, seek motion vector [Δ x, Δ y] with minimum SAD by calculating SAD ^TBe shown below:

$\mathrm{SAD}(\mathrm{\&Delta;x},\mathrm{\&Delta;y})=\frac{1}{\mathrm{mn}}\underset{(x,y)\&Element;W}{\mathrm{\&Sigma;}}|I(x,y,k)-I(x+\mathrm{\&Delta;x},y+\mathrm{\&Delta;y},k+1)|;$

${[\mathrm{\&Delta;x},\mathrm{\&Delta;y}]}^T=\arg \underset{(\mathrm{\&Delta;x},\mathrm{Xy})}{\min }\mathrm{SAD}(\mathrm{\&Delta;x},\mathrm{\&Delta;y}).$

Here, (x, y k) are the prior image frame data to I, and (x+ Δ x, y+ Δ y is k+1) for producing motion excursion [Δ x, Δ y] for I ^TAfter the current frame image data, wherein piecemeal W is of a size of m*n.

Above distortion is judged index just with the BTC boil down to example explanation compressed feedback of 2*4 piecemeal, and for other block sizes, and other compression algorithm is as the compression based on decorrelation, and as the DCT territory, the compression of wavelet field is revised a little too and is suitable for.

Preferably, in above-mentioned vedio data decompression method, further comprising the steps of:

When the distortion flag bit has been, utilize code stream block to recover the vedio data of current block, when estimation, the current frame data that utilizes current block and the corresponding decompression block data with better quality estimate to obtain the motion vector of current block by block matching motion.

Fig. 3 shows video image data compression apparatus module figure according to an embodiment of the invention, comprising:

Coding module 110 is used for vedio data is carried out piecemeal, and the current block of vedio data is compressed, and obtains code stream block;

Decoder module 120 is used for code stream block is decoded, and obtains the data reconstruction of current block;

Evaluation module 130, be used to calculate data reconstruction with respect to the compression of current block before the distortion level of data, and will reflect that the distortion flag bit of distortion level stores in the code stream block.

Fig. 4 shows vedio data decompressing device module map according to an embodiment of the invention, comprising:

Parsing module 210 is used for obtaining the distortion flag bit that reflects code stream block compression artefacts degree from code stream block;

Rebuilding module 220 if be used for the distortion flag bit for poor, when estimation, utilizes the motion vector interpolation estimation of the adjacent block of current block to obtain the motion vector of current block.

In the present embodiment, by obtaining the distortion flag bit in the code stream block, obtain the compression quality information of each blocks of data in the former frame, in application, just can adjust its follow-up video processnig algorithms in real time according to it, such as when the estimation, if less these decompression block data of can directly utilizing of distortion are carried out estimation, otherwise the motion vector that utilizes contiguous block estimates to obtain the motion vector of current block by interpolation.

Two frame data utilization block matching algorithm carried out the hardware schematic diagram of estimation of motion vectors before and after Fig. 5 showed and utilizes according to an embodiment of the invention, wherein the former frame view data adopts the compression of band feedback, and compression divides block size to divide block size identical with block matching algorithm.Near the piece vector of doing exercises reliably estimated to obtain the motion vector of current block by interpolation if former frame corresponding blocks compression artefacts flag bit for poor, was then selected, otherwise the motion vector that adopts piece to mate.

Preferably, in above-mentioned vedio data decompressing device, also comprise:

Decompressing device, be used for when the distortion flag bit has been, utilize code stream block to recover the vedio data of current block, when estimation, the current frame data that utilizes current block and the corresponding decompression block data with better quality obtain the motion vector of current block by the block matching motion estimation.

In the above embodiment of the present invention, consider the compression quality difference of compression algorithm to different pieces of information piece in the same image, in the encoding compression process, adopted feedback element.Its concrete thinking is exactly at coding side, after a certain blocks of data is finished encoding compression, then this code stream block is decoded then, calculate compression artefacts, quantize the compression algorithm that adopted distortion level with 1-2bit, and it is added in the corresponding code stream block current data block.In decoding end, when code stream block is decompressed, also obtain the distortion level of this data block lossy compression method, adjust follow-up video processnig algorithms according to it then.

Obviously, those skilled in the art should be understood that, above-mentioned each module of the present invention or each step can realize with the general calculation device, they can concentrate on the single calculation element, perhaps be distributed on the network that a plurality of calculation element forms, alternatively, they can be realized with the executable program code of calculation element, thereby, they can be stored in the storage device and carry out by calculation element, perhaps they are made into each integrated circuit modules respectively, perhaps a plurality of modules in them or step are made into the single integrated circuit module and realize.Like this, the present invention is not restricted to any specific hardware and software combination.

The above is the preferred embodiments of the present invention only, is not limited to the present invention, and for a person skilled in the art, the present invention can have various changes and variation.Within the spirit and principles in the present invention all, any modification of being done, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (9)

1. a video image data compression method is characterized in that, may further comprise the steps:

Described vedio data is carried out piecemeal;

The employing compression algorithm is compressed the current block of described vedio data, obtains code stream block;

Adopt the decoding algorithm corresponding that described code stream block is decoded, obtain the data reconstruction of described current block with described compression algorithm;

Calculate the distortion level of described data reconstruction, and will reflect that the distortion flag bit of described distortion level stores in the described code stream block with respect to the data of the preceding described current block of compression.

2. video image data compression method according to claim 1, it is characterized in that, described compression algorithm is following any compression algorithm: BTC algorithm, JPEG Coding Compression Algorithm, and based on one dimension or two-dimentional piecemeal compression algorithm, scalar quantization algorithm, Vector Quantization algorithm and the grid coding quantization algorithm of wavelet transformation.

3. video image data compression method according to claim 1 is characterized in that, calculate described data reconstruction with respect to the compression of described current block before the distortion level of data specifically comprise:

Calculate the error or the coefficient correlation of the preceding data of compression of described data reconstruction and described current block;

The threshold value of described error or described coefficient correlation and setting is compared, obtain the distortion level of the packed data of described current block, and represent described distortion level with limited bit.

4. video image data compression method according to claim 3 is characterized in that, described error is worst error or mean error or mean square error.

5. a vedio data decompression method is characterized in that, may further comprise the steps: the distortion flag bit that obtains the described code stream block compression losses of reflection from code stream block;

If described distortion flag bit is for poor, when estimation, utilize the motion vector interpolation estimation of the adjacent block of current block to obtain the motion vector of described current block.

6. vedio data decompression method according to claim 5 is characterized in that, and is further comprising the steps of:

If described distortion flag bit is for well, utilize described code stream block to recover the vedio data of described current block, when estimation, the current frame data that utilizes described current block and the corresponding decompression block data with better quality estimate to obtain the motion vector of described current block by block matching motion.

7. a video image data compression device is characterized in that, comprising:

Coding module is used for described vedio data is carried out piecemeal, and the current block of described vedio data is compressed, and obtains code stream block;

Decoder module is used for described code stream block is decoded, and obtains the data reconstruction of described current block;

Evaluation module, be used to calculate described data reconstruction with respect to the compression of described current block before the distortion level of data, and will reflect that the distortion flag bit of described distortion level stores in the described code stream block.

8. a vedio data decompressing device is characterized in that, comprising:

Parsing module is used for obtaining the distortion flag bit that reflects described code stream block compression artefacts degree from code stream block;

Rebuilding module if be used for described distortion flag bit for poor, when estimation, utilizes the motion vector interpolation estimation of the adjacent block of current block to obtain the motion vector of described current block.

9. vedio data decompressing device according to claim 8 is characterized in that, also comprises:

Decompression module, be used for when described distortion flag bit has been, utilize described code stream block to recover the vedio data of described current block, when estimation, the current frame data that utilizes described current block and the corresponding decompression block data with better quality estimate to obtain the motion vector of described current block by block matching motion.

Images