Mode searching and early termination of a video picture and fast compression of variable length symbols
View Patent ↗Mode searching and early termination of a video picture may include determining the cost value of a simple intra-mode prediction of the video picture; determining the cost value of an inter-mode prediction of the video picture; determining a threshold value based on a given quantization parameter (QP); comparing the cost value of the simple intra-mode prediction to the sum of the cost value of the inter-mode prediction and the threshold value; and performing a mode-prediction of the video picture in accordance with the result obtained. Fast compression of variable length symbols may include receiving a variable length symbol with one or more tokens; combining the one or more tokens in the variable length symbol to form a token combination; and determining a bit representation of the token combination in correspondence to a probability threshold associated with the token combination.
1. A method for mode searching and early termination of a video picture, comprising:
a) determining the cost value of a simple intra-mode prediction of the video picture;
b) determining the cost value of an inter-mode prediction of the video picture;
c) determining a threshold value based on a given quantization parameter (QP);
d) comparing the cost value of the simple intra-mode prediction to the sum of the cost value of the inter-mode prediction and the threshold value; and
e) performing a mode-prediction of the video picture in accordance with the result obtained in d).
2. The method of claim 1 , wherein e) includes performing an inter-mode prediction if the cost value of the simple intra-mode prediction is greater than the sum of the cost value of the inter-mode prediction and the threshold value.
3. The method of claim 1 , wherein e) includes performing an intra-mode prediction if the cost value of the simple intra-mode prediction is lower than the sum of the cost value of the inter-mode prediction and the threshold value.
4. The method of claim 1 , wherein the simple intra-mode prediction for a given section does not wait for an intra prediction result of neighbor section.
5. The method of claim 1 , wherein the simple intra-mode search uses a simplified 16-bit quantization process.
6. The method of claim 1 , wherein the simplified intra-mode search includes a search for the best intra prediction for input pixels of a given macroblock in a section,
using the best intra prediction to calculate a prediction error,
converting the prediction error into residual DCT coefficients using DCT transforms,
constructing a locally decoded macroblock
using the locally decoded macroblock to do intra prediction for one or more neighbor macroblocks of the given macroblock.
7. The method of claim 6 wherein a simplified 16 bit quantization is used to obtain quantized DCT coefficients.
8. The method of claim 7 wherein the locally decoded macroblock is constructed from pixel predictors and the quantized DCT coefficients.
9. The method of claim 7 , wherein the quantized DCT coefficients are not for entropy encoding.
10. The method of claim 6 wherein locally decoded macroblock is used to do intra prediction for its right and lower neighbor macroblocks.