Title :
Fast Intra Prediction Mode Decision by Adaptively Selecting Fewer Number of Modes
Author :
Jamil-Ur-Rehman ; Ye, Zhang ; Kumarayapa, Ajith
Author_Institution :
Harbin Inst. of Technol., Harbin
Abstract :
Enhanced Intra and Inter prediction techniques are the key factors to the success of H.264. There are a total of nine optional prediction modes for each 4 x 4 luma block. H.264 adapts Rate Distortion Optimization (RDO) technique to get high coding rate and visual quality. Because RDO is too much exhaustive, several attempts have been made to explore fast algorithms. The choice of intra prediction mode for each 4 x 4 block must be signaled to the decoder and this could potentially require a large number of bits. However, intra modes for neighbouring 4 x 4 blocks are often correlated. To take advantage of this correlation, predictive coding is used to signal 4 x 4 intra modes. But still large number of bits is required to signal nine intra prediction modes. We adaptively selected only five modes; selection of modes depends on the position of 4 x 4 block in the slice or frame. After experimental results we have come to know that the increase in the number of bits for encoding residual coefficients is approximately same as decrease in the number of bits required to signal intra prediction modes, providing same PSNR (peak signal to noise ratio). By using proposed technique, computational speed increased by ap45% without significant performance degradation.
Keywords :
optimisation; rate distortion theory; video coding; H.264; intra prediction mode decision; luma block; peak signal to noise ratio; predictive coding; rate distortion optimization; Bit rate; Cybernetics; Decoding; Degradation; Electronic mail; Encoding; Machine learning; PSNR; Predictive coding; Rate-distortion; H.264; Intra prediction; Mode decision; Signaling mode numbers;
Conference_Titel :
Machine Learning and Cybernetics, 2007 International Conference on
Conference_Location :
Hong Kong
Print_ISBN :
978-1-4244-0973-0
Electronic_ISBN :
978-1-4244-0973-0
DOI :
10.1109/ICMLC.2007.4370544