CN100586184C - Infra-frame prediction method - Google Patents

Abstract

The invention relates to an intra-frame prediction method, comprising the following steps that: a 4x4 pixel luma block to be predicted is chosen to be as a current block; whether the current block is the center block is judged, if yes, energy functions of the residual error between prediction blocks of the current block and the current block in all predicting modes and a candidate predicting mode is determined according to the energy functions in different predicting modes, otherwise, a usable predicting mode is taken as the candidate predicting mode; a rate-distortion cost parameter in the candidate predicting mode is calculated, the candidate predicting mode with the minimal rate-distortion cost parameter is taken as the optimal predicting mode of the current block. The invention determines the candidate predicting mode according to the residual error energy function and determines the optimal predicting mode by calculating the rate-distortion cost parameter of the candidate predicting mode so as to decrease the number of the predicting modes which need to calculate the rate-distortion cost parameter, thereby reducing the amount of computation.

Description

Intra-frame prediction method

Technical field

The present invention relates to a kind of intra-frame prediction method, especially a kind of intra-frame prediction method that can effectively improve arithmetic speed.

Background technology

H.264/AVC, a new generation's video encoding standard has the good network performance, is applicable to mutual and nonreciprocal applied environment, has just received very big concern since issue, and its significant advantage is the compression efficiency height.Fig. 1 shows the H.264/AVC encoder frame figure of standard, and its course of work can be divided into forward path and rebuild passage according to data flow direction.Present frame F _nCoding, be that the macro block of original image 16 * 16 pixels is encoded, macroblock coding is divided into intraframe coding and interframe encode, when intraframe coding and interframe encode, predicted macroblock P is obtained by reconstruction frames.In intra-frame encoding mode, P by the coded macroblocks in the present frame through the decoding, reconstruct, by reconstruct macro block uF ' _nPrediction obtains, and under interframe encoding mode, P is by one or more reference frame F ' _N-1Motion compensated prediction obtains.Predicted macroblock P and current macro F _nDifference be residual error macro block D _n, D _nObtain a string transformation parameter X after conversion, quantification, parameter X need be carried out the processing of two aspects: the one, reorder and the entropy coding processing, and whole process does not have feedback component, so be called forward path; The 2nd, inverse quantization and inversion process produce macro block D ' _n, D ' _nObtain reconstruct macro block uF ' with predicted macroblock P addition _n, reconstruct macro block uF ' _nPass through the reference frame F ' that a series of processing obtain rebuilding again _n, be used for the estimation of next frame, so be called the reconstruction passage.H.264/AVC the raising of performance is inseparable with some new technologies that it adopts, as adopt intra-frame prediction method, conversion based on spatial domain adopt integral discrete cosine transform (Discrete Cosine Transform is hereinafter to be referred as DCT), motion estimation/motion compensation based on 4 * 4 block of pixels, rebuild passage and adopt loop filtering, new entropy coding method etc.

Be the key factor that improves of performance H.264/AVC wherein based on the infra-frame prediction of spatial domain, infra-frame prediction utilizes the spatial coherence of image, according to the information of the information prediction current block of the adjacent pixel blocks of decoding and reconstituting, obtain the prediction piece of current block, then the residual error of the prediction piece of current block and current block is carried out conversion, quantification, coding, in order to represent current block preferably, H.264/AVC adopt rate-distortion optimization (RateDistortion Optization is hereinafter to be referred as RDO) technology to come the optimized encoding quality and minimize bits of coded.

H.264/AVC the macro block with 16 * 16 pixels is a unit in the standard, a macro block comprises the chrominance block of a luminance block and two correspondences, wherein luminance block be 1 16 * 16 or 16 4 * 4, chrominance block is 8 * 8.

Luminance block for 4 * 4, infra-frame prediction have 9 kinds of predictive modes, and these predictive modes are:

Pattern 0: vertical prediction (vertical prediction) pattern

Pattern 1: horizontal forecast (horizontal prediction) pattern

Pattern 2:DC predicts (DC prediction) pattern

Mode 3: 45 degree direction prediction (diagonal down/left prediction) patterns

Pattern 4:135 degree direction prediction (diagonal down/right prediction) pattern

Pattern 5:112.5 degree direction prediction (vertical-right prediction) pattern

Pattern 6:157.5 degree direction prediction (horizontal-down prediction) pattern

Mode 7: 67.5 degree direction prediction (vertical-left prediction) patterns

Pattern 8:22.5 degree direction prediction (horizontal-up prediction) pattern

Wherein, except the DC predictive mode, remaining 8 kinds of predictive modes are called directional prediction modes, and Fig. 2 shows the graph of a relation of reference pixel and current block pixel to be predicted, and wherein capitalization is represented is reference pixel, and what lowercase was represented is the current block pixel.Numeral among Fig. 3 has been indicated the sensing of all directions predictive modes.

Luminance block for 16 * 16, infra-frame prediction have 4 kinds of predictive modes, and these predictive modes are:

Vertical prediction (vertical prediction) pattern

Horizontal forecast (horizontal prediction) pattern

DC predicts (DC prediction) pattern

Dull and stereotyped prediction (plane prediction) pattern

Chrominance block for 8 * 8, infra-frame prediction have 4 kinds of predictive modes, and these predictive modes are:

DC predicts (DC prediction) pattern

Horizontal forecast (horizontal prediction) pattern

Vertical prediction (vertical prediction) pattern

Dull and stereotyped prediction (plane prediction) pattern

From above-mentioned analysis as can be seen: represent that quantity, the M16 of the predictive mode of 4 * 4 luminance block represent the quantity of the predictive mode of 16 * 16 luminance block if M8 represents quantity, the M4 of the predictive mode of chrominance block, the number of combinations of luminance block and chroma block prediction modes is M8 * (M4 * 16+M16)=592 in macro block, promptly, for the optimization model of the infra-frame prediction of determining a macro block, encoder need carry out the calculating of 592 RDO.Therefore, H.264/AVC the raising of compression ratio is the cost that increases to computation complexity, is difficult to real-time application.

Summary of the invention

The objective of the invention is defective, a kind of intra-frame prediction method is provided,, improve the speed of intraframe predictive coding, help the real-time application of encoder in order to solve the existing high problem of intra-frame prediction method computation complexity at prior art.

The present invention provides following technical scheme by embodiment:

A kind of intra-frame prediction method comprises:

Step 1: choose 4 * 4 pixel intensity pieces to be predicted as current block;

Step 2: judge whether described current block is central block, if, execution in step 3, otherwise, execution in step 4;

Step 3: calculate the energy function of the residual error of the prediction piece of described current block under all predictive modes and described current block, and determine candidate's predictive mode, execution in step 5 according to the energy function under the different predictive modes;

Wherein said step 3 specifically comprises:

Step 31: the prediction piece that calculates the described current block under the different predictive modes;

Step 32: prediction piece and described current block according to the described current block under the different predictive modes obtain the prediction piece of the described current block under the different predictive modes and the residual error of described current block;

Step 33: the residual error under the different predictive modes is carried out the energy function that discrete cosine transform obtains the residual error under the different predictive modes;

Wherein the computing formula of energy function is:

$E=\underset{y=1}{\overset{4}{\mathrm{\Σ}}}\underset{x=1}{\overset{4}{\mathrm{\Σ}}}w(y,x)\mathrm{DCT}(y-1,x-1)$

Wherein, $w(y,x)=\left\{\begin{array}{cc}0.2& (y=1,x=1),(y=1,x=2),(y=2,x=1)\\ 0.4& (y=1,x=3),(y=2,x=2),(y=3,x=1)\\ 0.6& (y=1,x=4),(y=2,x=3),(y=3,x=2),(y=4,x=1)\\ 0.8& (y=2,x=4),(y=3,x=3),(y=4,x=2)\\ 1& (y=3,x=4),(y=4,x=3),(y=4,x=4)\end{array}\right.$

$\mathrm{DCT}=\left(\begin{array}{c}\left[\begin{array}{cccc}1& 1& 1& 1\\ 2& 1& -1& -2\\ 1& -1& -1& 1\\ 1& -2& 2& -1\end{array}\right]\mathrm{<figure-callout\; id="1"\; label="X"\; filenames="C2008100567760016H2.png"\; state="\{\{state\}\}">X</figure-callout>}\left[\begin{array}{cccc}1& 2& 1& 1\\ 1& 1& -1& -2\\ 1& -1& -1& 2\\ 1& -2& 1& -1\end{array}\right]\end{array}\right)\&CircleTimes;\left[\begin{array}{cccc}{a}^2& \frac{\mathrm{ab}}{2}& a^2& \frac{\mathrm{ab}}{2}\\ \frac{\mathrm{ab}}{2}& \frac{b^2}{4}& \frac{\mathrm{ab}}{2}& \frac{b^2}{4}\\ a^2& \frac{\mathrm{ab}}{2}& a^2& \frac{\mathrm{ab}}{2}\\ \frac{\mathrm{ab}}{2}& \frac{b^2}{4}& \frac{\mathrm{ab}}{2}& \frac{b^2}{4}\end{array}\right]$

Y is the coordinate of the Width of described current block, and x is the coordinate of the short transverse of described current block, and X is the prediction piece of described current block and the residual error of described current block, DCT (y, x) be described current block coordinate for (y, DCT coefficient x), $a=\frac{1}{2},$ $b=\sqrt{\frac{2}{5}};$

Step 34: with the energy function of the residual error under the different predictive modes according to energy value sequence arrangement from small to large, obtain the first less energy of energy value, second energy, the 3rd energy and the 4th energy successively, and obtain corresponding to described first energy first predictive mode, corresponding to second predictive mode of described second energy, corresponding to the 3rd predictive mode of described the 3rd energy with corresponding to the 4th predictive mode of described the 4th energy;

Step 35: determine candidate's predictive mode according to described first energy, second energy, the 3rd energy and the 4th energy and described first predictive mode, second predictive mode, the 3rd predictive mode and the 4th predictive mode;

Step 4: with available predictive mode as candidate's predictive mode, execution in step 5;

Step 5: the rate distortion costs parameter of calculated candidate predictive mode, candidate's predictive mode of choosing rate distortion costs parameter minimum is the optimal prediction modes of described current block.

The embodiment of the invention is selected candidate's predictive mode of 4 * 4 pixel intensity pieces according to energy function, rate distortion costs parameter by the calculated candidate predictive mode obtains optimal prediction modes, rather than obtain optimal prediction modes according to the rate distortion costs parameter of all predictive modes, because reducing of predictive mode quantity, can slip distortion cost CALCULATION OF PARAMETERS amount, improve arithmetic speed, be applicable to real-time coding.

Description of drawings

Fig. 1 is the H.264/AVC encoder frame figure of standard of prior art;

Fig. 2 is the H.264/AVC schematic diagram of reference pixel and current block pixel relationship to be predicted in the standard of prior art;

Fig. 3 is the H.264/AVC sensing figure of directional prediction modes in the standard of prior art;

Fig. 4 is intra-frame prediction method embodiment one flow chart of the present invention;

Fig. 5 is intra-frame prediction method embodiment two flow charts of the present invention.

Embodiment

Further specify technical scheme of the present invention below in conjunction with the drawings and specific embodiments.

Fig. 4 is intra-frame prediction method embodiment one flow chart of the present invention, and this intra-frame prediction method comprises:

Step 41: choose 4 * 4 pixel intensity pieces to be predicted as current block;

Step 42: judge whether described current block is central block, if, execution in step 43, otherwise, execution in step 44;

Step 43: calculate the energy function of the residual error of the prediction piece of described current block under all predictive modes and described current block, and determine candidate's predictive mode, execution in step 45 according to the energy function under the different predictive modes;

Step 44: with available predictive mode as candidate's predictive mode, execution in step 45;

Step 45: the rate distortion costs parameter of calculated candidate predictive mode, candidate's predictive mode of choosing rate distortion costs parameter minimum is the optimal prediction modes of described current block.

Fig. 5 is intra-frame prediction method embodiment two flow charts of the present invention, and this intra-frame prediction method comprises:

Step 501: with one 16 * 16 macroblock partitions is 16 4 * 4 luminance block, execution in step 502.H.264/AVC the macro block with 16 * 16 pixels is a unit in the standard, a macro block comprises a luminance block and two chrominance block, luminance block be 1 16 * 16 or 16 4 * 4, chrominance block is 8 * 8, the pixel value of current luminance block or chrominance block use last piece adjacent and adjacent left piece with current block the pixel value of decoding and reconstituting according to multiple predictive mode predictive coding.Because H.264/AVC the conversion in the standard is a unit with 4 * 4, and 4 * 4 occupy very big ratio in predictive coding, therefore, present embodiment improves at the predictive coding of 4 * 4 luminance block, will not change the predictive coding method of 16 * 16 luminance block and 8 * 8 chrominance block.

Step 502: 4 * 4 luminance block of choosing a coding to be predicted are as current block, execution in step 503.The predictive mode of the intra-frame prediction method of 4 * 4 luminance block comprises 8 kinds of direction predictions and a kind of DC prediction.The selection of adopting the RDO technology to carry out the predictive mode of luminance block in traditional H.264/AVC standard comprises the steps:

(1) for 4 * 4 luminance block, calculate 9 kinds of rate distortion costs parameters under the predictive mode respectively, obtain the minimum value of rate distortion costs parameter and corresponding to the predictive mode of rate distortion costs parameter minimum.The computing formula of rate distortion costs parameter (RDCost) is: RDCost=SSD+ λ * bitrate, and wherein RDCost is the rate distortion costs parameter; SSD is the quadratic sum of all elements of the difference block of current block and reconstructed blocks, i.e. reconstruct residual error square; λ is the function of quantization parameter (QP), λ=0.85 * 2 ^(QP-12)/3Bitrate is the code check behind the entropy coding, promptly forms the bit number of code stream after the predictive coding.

(2) obtain the minimum rate distortion costs parameter of 16 4 * 4 luminance block according to said method, and with this 16 minimum rate distortion costs parameters addition, with after the addition and as the rate distortion costs parameter of 4 * 4 luminance block.

(3) for 16 * 16 luminance block, calculate respectively all elements after the hasmard conversion of residual error of 4 kinds of current blocks under the predictive mode and prediction piece and, be residual error hasmard conversion and value (STAD), calculate the rate distortion costs parameter of the minimum predictive mode of STAD value, with the rate distortion costs parameter of this rate distortion costs parameter as 16 * 16 luminance block.

(4) the rate distortion costs parameter that obtains of (2), (3) is relatively selected the predictive mode of the predictive mode of minimum rate distortion costs parameter correspondence as luminance block in the frame.

From the formula of the rate distortion costs parameter of above-mentioned (1) as can be seen, every kind of predictive mode to current 4 * 4 luminance block all will carry out precoding, computation complexity is bigger, in order to reduce computation complexity, can simplify the calculating of RDO or the predictive mode that minimizing is used for calculating RDO, for the complexity that reduces intraframe coding keeps the efficient of intraframe coding simultaneously preferably, present embodiment adopts and reduces the method that is used for calculating the RDO predictive mode.

Step 503: judging whether current block is central block, is execution in step 504 then, otherwise, execution in step 507.Because adopt from left to right in the video coding, order from top to bottom carries out decoding and reconstituting, when if left piece adjacent with current block and adjacent last piece have all been decoded, be called left piece and last piece available, when left piece decoding and reconstituting, it is available to be called left piece, and when last piece during decoding and reconstituting, it is available to be called piece, when both did not all decode, it was all unavailable to be called both.But when left piece and last piece equal time spent, current block is central block, at this moment, corresponding to current block 9 kinds of predictive modes is arranged; When current block is not central block, predictive mode corresponding to current block is called the available predictions pattern, 9 kinds of available predictions pattern less thaies, but as has only the left piece time spent, the available predictions pattern is with horizontal forecast pattern (pattern 1), DC predictive mode (pattern 2), 22.5 degree directional prediction modes (pattern 8), but had only the last piece time spent, the available predictions pattern is vertical prediction pattern (pattern 0), DC predictive mode (pattern 2), 45 degree directional prediction modes (mode 3), 67.5 degree direction predictions (mode 7).

Step 504: calculate the energy function of residual error of current block and prediction piece under all predictive modes, execution in step 505.From above-mentioned analysis as can be known, if when current block is central block, its available predictive mode has 9 kinds, in order to reduce the operand that RDO calculates, adopts the method that reduces predictive mode.Because most of zone of nature or synthetic image is made up of low frequency, and most of high-frequency information can be used 0 run-length encoding, and high frequency can reflect the complexity of the texture of an image roughly.In actual applications, predict the information of current block according to the grain direction of current block with the information of the adjacent block of decoding, then corresponding residual error is carried out the calculating of energy function, the defined energy function of the present invention is to the mapping of the energy space of one dimension from the frequency domain space of multidimensional, the high fdrequency component of the dct transform of texture residual block has reflected the detailed information of texture residual error, detailed information for outstanding texture residual error, the present invention distributes the higher weights (influence that w (y, x)) can amplify high frequency coefficient for high frequency.By the definition of DCT energy function of the present invention, texture residual block details abundant more (high-frequency information is many more), its DCT energy is big more.The present invention has utilized this feature of energy function just, select the several modes of texture residual block complexity less (energy is less, high-frequency information is less) to carry out the calculating of RDO, thereby can reach to reduce and be used for carrying out the pattern quantity that RDO calculates, the final purpose that realizes accelerating intraframe coding speed.Wherein the computing formula of energy function is:

$E_{\mathrm{DCT}}=\underset{y=1}{\overset{\mathrm{\&omega;}}{\mathrm{\&Sigma;}}}\underset{x=1}{\overset{h}{\mathrm{\&Sigma;}}}w(y,x)\mathrm{DCT}(y-1,x-1)$

Here E _DCTBe the energy calculating at the dct transform of prediction residual, ω and h are the width and the height of current block, and (y is to be positioned at (y, DCT coefficient x) x) to DCT, for 4 * 4 luminance block H.264/AVC, weights function w in the following formula (y, x) (x=1,2 ... h, y=1,2 ..., ω), be defined as follows:

$w(y,x)=\left\{\begin{array}{cc}0.2& (y=1,x=1),(y=1,x=2),(y=2,x=1)\\ 0.4& (y=1,x=3),(y=2,x=2),(y=3,x=1)\\ 0.6& (y=1,x=4),(y=2,x=3),(y=3,x=2),(y=4,x=1)\\ 0.8& (y=2,x=4),(y=3,x=3),(y=4,x=2)\\ 1& (y=3,x=4),(y=4,x=3),(y=4,x=4)\end{array}\right.$

Wherein, ω=4, h=4

Integer dct transform in the dct transform employing standard in the following formula, formula is as follows:

$\mathrm{DCT}=\left(\begin{array}{c}\left[\begin{array}{cccc}1& 1& 1& 1\\ 2& 1& -1& -2\\ 1& -1& -1& 1\\ 1& -2& 2& -1\end{array}\right]\mathrm{<figure-callout\; id="1"\; label="X"\; filenames="C2008100567760016H2.png"\; state="\{\{state\}\}">X</figure-callout>}\left[\begin{array}{cccc}1& 2& 1& 1\\ 1& 1& -1& -2\\ 1& -1& -1& 2\\ 1& -2& 1& -1\end{array}\right]\end{array}\right)\&CircleTimes;\left[\begin{array}{cccc}{a}^2& \frac{\mathrm{ab}}{2}& a^2& \frac{\mathrm{ab}}{2}\\ \frac{\mathrm{ab}}{2}& \frac{b^2}{4}& \frac{\mathrm{ab}}{2}& \frac{b^2}{4}\\ a^2& \frac{\mathrm{ab}}{2}& a^2& \frac{\mathrm{ab}}{2}\\ \frac{\mathrm{ab}}{2}& \frac{b^2}{4}& \frac{\mathrm{ab}}{2}& \frac{b^2}{4}\end{array}\right]$

Wherein, $a=\frac{1}{2},$ $b=\sqrt{\frac{2}{5}},$ X is current block and the residual error of predicting piece.

Step 505: with the energy function value of the reconstruct residual error under the different predictive modes by from small to large sequence arrangement, choose minimum four energy value first energy (E1), second energy (E2), the 3rd energy (E3), the 4th energy (E4) and corresponding first predictive mode (m1), second predictive mode (m2), the 3rd predictive mode (m3), the 4th predictive mode (m4), execution in step 506.

Step 506:, from above-mentioned four kinds of predictive mode m1, m2, m3, m4, determine candidate's predictive mode, execution in step 508 according to following screening principle.Described screening principle comprises:

Step 5061: whether judge E2 greater than α * E1, if, execution in step 5062; Otherwise, execution in step 5063.

Step 5062: candidate's predictive mode is m1.

Step 5063: whether judge E3 greater than α * E1, if, execution in step 5064, otherwise execution in step 5065.

Step 5064: candidate's predictive mode is m1 and m2.

Step 5065: whether judge E4 greater than α * E1, if, execution in step 5066, otherwise execution in step 5067.

Step 5066: candidate's predictive mode is m1, m2 and m3.

Step 5067: whether judge E4 less than β * E1, if, execution in step 5068, otherwise, execution in step 5069.

Step 5068: candidate's predictive mode is m1 and m2.

Step 5069: candidate's predictive mode is m1, m2, m3 and m4.

Above-mentioned α, β are constants, through a large amount of tests, get α=1.4 and β=1.05 o'clock coding best results.

Step 507: with available predictive mode as candidate's predictive mode, execution in step 508.But when having only the left piece time spent adjacent with current block, the available predictions pattern is pattern 1, pattern 2, pattern 8; But when having only the last piece time spent adjacent with current block, the available predictions pattern is pattern 0, pattern 2, mode 3, mode 7; When the left piece adjacent with current block and last piece when all unavailable, candidate's predictive mode is the DC predictive mode.

Step 508: each candidate's predictive mode of current block is carried out RDO calculate, candidate's predictive mode of choosing rate distortion costs parameters R DCost minimum is the optimal prediction modes of current block, execution in step 509.But be the adjacent left piece and adjacent the going up the equal time spent of piece of current block, carry out RDO according to candidate's predictive mode that energy function is determined and calculate that candidate's predictive mode of the rate distortion costs parameter minimum that obtains is an optimal prediction modes; But the adjacent left piece of current block or adjacent going up one of the piece time spent are carried out RDO to available predictive mode and are calculated, and the available predictions pattern of the rate distortion costs parameter minimum that obtains is an optimal prediction modes; When the adjacent left piece of current block and adjacent last piece were all unavailable, the DC predictive mode was an optimal prediction modes.

Step 509: judge whether that 16 luminance block all carried out aforementioned calculation as current block, if, execution in step 510, otherwise repeated execution of steps 502 is all finished predictive coding up to 4 * 4 all luminance block.

Step 510: with the minimum rate distortion costs parameter addition of 16 4 * 4 luminance block, and with after the addition and as the rate distortion costs parameters R DCost4 of 4 * 4 luminance block, execution in step 512.

Step 511: the rate distortion costs parameter of calculating 4 kinds of predictive modes of 16 * 16 luminance block, obtain minimum rate distortion costs parameter and corresponding to the predictive mode of minimum rate distortion costs parameter, wherein Zui Xiao rate distortion costs parameter is the rate distortion costs parameters R DCost16 of 16 * 16 luminance block, predictive mode corresponding to minimum rate distortion costs parameter is the optimal prediction modes of 16 * 16 luminance block, execution in step 512.

Step 512: compare the rate distortion costs parameters R Dcost4 of 4 * 4 luminance block and the rate distortion costs parameters R Dcost16 of 16 * 16 luminance block, if RDcost4＜RDcost16, execution in step 513, otherwise, execution in step 514.

Step 513: choose corresponding to the optimal prediction modes of each 4 * 4 luminance block optimal prediction modes as the luminance block of 16 * 16 macro blocks.

Step 514: choose corresponding to the optimal prediction modes of 16 * 16 luminance block of RDcost16 optimal prediction modes as the luminance block of 16 * 16 macro blocks.

Present embodiment, intra-frame prediction method to 4 * 4 luminance block in the standard H.264/AVC improves, determine candidate's predictive mode with the energy function that the dct transform of residual error obtains, candidate's predictive mode is carried out RDO to be calculated, rather than all predictive modes are carried out RDO calculate, owing to carry out the predictive mode quantity minimizing that RDO calculates, can reduce the operand of RDO effectively, help being applied to real-time coding.

It should be noted that at last: above embodiment only in order to technical scheme of the present invention to be described, is not intended to limit; Although with reference to previous embodiment the present invention is had been described in detail, those of ordinary skill in the art is to be understood that: it still can be made amendment to the technical scheme that aforementioned each embodiment put down in writing, and perhaps part technical characterictic wherein is equal to replacement; And these modifications or replacement do not make the essence of appropriate technical solution break away from the spirit and scope of various embodiments of the present invention technical scheme.

Claims (5)

1, a kind of intra-frame prediction method is characterized in that, comprising:

Step 1: choose 4 * 4 pixel intensity pieces to be predicted as current block;

Step 2: judge whether described current block is central block, if, execution in step 3, otherwise, execution in step 4;

Step 3: calculate the energy function of the residual error of the prediction piece of described current block under all predictive modes and described current block, and determine candidate's predictive mode, execution in step 5 according to the energy function under the different predictive modes;

Wherein said step 3 specifically comprises:

Step 31: the prediction piece that calculates the described current block under the different predictive modes;

Step 32: prediction piece and described current block according to the described current block under the different predictive modes obtain the prediction piece of the described current block under the different predictive modes and the residual error of described current block;

Step 33: the residual error under the different predictive modes is carried out the energy function that discrete cosine transform obtains the residual error under the different predictive modes;

Wherein the computing formula of energy function is:

$E=\underset{y=1}{\overset{4}{\mathrm{\&Sigma;}}}\underset{x=1}{\overset{4}{\mathrm{\&Sigma;}}}w(y,x)\mathrm{DCT}(y-1,x-1)$

Wherein, $w(y,x)=\left\{\begin{array}{cc}0.2& (y=1,x=1),(y=1,x=2),(y=2,x=1)\\ 0.4& (y=1,x=3),(y=2,x=2),(y=3,x=1)\\ 0.6& (y=1,x=4),(y=2,x=3),(y=3,x=2),(y=4,x=1)\\ 0.8& (y=2,x=4),(y=3,x=3),(y=4,x=2)\\ 1& (y=3,x=4),(y=4,x=3),(y=4,x=4)\end{array}\right.$

$\mathrm{DCT}=\left(\left[\begin{array}{cccc}1& 1& 1& 1\\ 2& 1& -1& -2\\ 1& -1& -1& 1\\ 1& -2& 2& -1\end{array}\right]X\left[\begin{array}{cccc}1& 2& 1& 1\\ 1& 1& -1& -2\\ 1& -1& -1& 2\\ 1& -2& 1& -1\end{array}\right]\right)\&CircleTimes;\left[\begin{array}{cccc}{a}^2& \frac{\mathrm{ab}}{2}& a^2& \frac{\mathrm{ab}}{2}\\ \frac{\mathrm{ab}}{2}& \frac{b^2}{4}& \frac{\mathrm{ab}}{2}& \frac{b^2}{4}\\ a^2& \frac{\mathrm{ab}}{2}& a^2& \frac{\mathrm{ab}}{2}\\ \frac{\mathrm{ab}}{2}& \frac{b^2}{4}& \frac{\mathrm{ab}}{2}& \frac{b^2}{4}\end{array}\right]$

Y is the coordinate of the Width of described current block, and x is the coordinate of the short transverse of described current block, and X is the prediction piece of described current block and the residual error of described current block, DCT (y, x) be described current block coordinate for (y, DCT coefficient x), $a=\frac{1}{2},b=\sqrt{\frac{2}{5}};$

Step 34: with the energy function of the residual error under the different predictive modes according to energy value sequence arrangement from small to large, obtain the first less energy of energy value, second energy, the 3rd energy and the 4th energy successively, and obtain corresponding to described first energy first predictive mode, corresponding to second predictive mode of described second energy, corresponding to the 3rd predictive mode of described the 3rd energy with corresponding to the 4th predictive mode of described the 4th energy;

Step 35: determine candidate's predictive mode according to described first energy, second energy, the 3rd energy and the 4th energy and described first predictive mode, second predictive mode, the 3rd predictive mode and the 4th predictive mode;

Step 4: with available predictive mode as candidate's predictive mode, execution in step 5;

Step 5: the rate distortion costs parameter of calculated candidate predictive mode, candidate's predictive mode of choosing rate distortion costs parameter minimum is the optimal prediction modes of described current block.

2, intra-frame prediction method according to claim 1, it is characterized in that, described step 2 is specially: the pixel intensity piece is pressed from left to right, the decoding of order from top to bottom, when the left pixel intensity piece adjacent with described current block and last pixel intensity piece had all been decoded, described current block was a central block.

3, intra-frame prediction method according to claim 1 is characterized in that, described step 35 specifically comprises:

Step 351: judge described second energy whether greater than 1.4 times of described first energy, if then described first predictive mode is candidate's predictive mode, otherwise, execution in step 352;

Step 352: judge described the 3rd energy whether greater than 1.4 times of described first energy, if then described first predictive mode and second predictive mode are candidate's predictive mode, otherwise, execution in step 353;

Step 353: judge described the 4th energy whether greater than 1.4 times of described first energy, if then described first predictive mode, second predictive mode and the 3rd predictive mode are candidate's predictive mode, otherwise, execution in step 354;

Step 354: judge that described the 4th energy is whether less than 1.05 times of described first energy, if, then described first predictive mode and second predictive mode are candidate's predictive mode, otherwise described first predictive mode, second predictive mode, the 3rd predictive mode and the 4th predictive mode are candidate's predictive mode.

4, intra-frame prediction method according to claim 1 is characterized in that, described step 4 specifically comprises:

When the last pixel intensity piece adjacent with described current block decoded, vertical prediction pattern, DC predictive mode, 45 degree directional prediction modes, 67.5 are spent directional prediction modes as candidate's predictive mode;

When the left pixel intensity piece adjacent with described current block decoded, horizontal forecast pattern, DC predictive mode, 22.5 are spent directional prediction modes as candidate's predictive mode;

When the last pixel intensity piece adjacent, when left pixel intensity piece is not all decoded with described current block, with the DC predictive mode as candidate's predictive mode.

5, intra-frame prediction method according to claim 1 is characterized in that, also comprises:

Step 6: with the rate distortion costs parameter sum of 16 4 * 4 pixel intensity piece optimal prediction modes rate distortion costs parameter as 4 * 4 pixel intensity pieces;

Step 7: calculate 16 * 16 pixel intensity pieces under different predictive modes residual error hasmard conversion and value;

Step 8: will have residual error hasmard conversion and the rate distortion costs parameter of predictive mode of value minimum as the rate distortion costs parameter of 16 * 16 pixel intensity pieces;

Step 9: whether the rate distortion costs parameter of judging described 4 * 4 pixel intensity pieces is less than the rate distortion costs parameter of described 16 * 16 pixel intensity pieces, if, will be corresponding to the optimal prediction modes of 4 * 4 pixel intensity pieces optimal prediction modes as the pixel intensity piece, otherwise, will be corresponding to the optimal prediction modes of 16 * 16 pixel intensity pieces optimal prediction modes as the pixel intensity piece.

Images