Title :
An improved inter-prediction motion estimation algorithm of AVS
Author :
Qiang, Wu ; Ying, Xu ; Kebin, Jia ; Xuwen, Li
Author_Institution :
Coll. of Electron. & Control Eng., Beijing Univ. of Technol., Beijing, China
Abstract :
Research on inter-prediction algorithms is an important respect in the real-time video codec technology. In the regarded studies, the prediction error can be reduced by incorporating high accuracy estimation and compensation algorithms. So the performance of compressed video can be improved. However, this brings a considerable cost for the complexity of encoder. The computation complexity is mainly due to the complexity of motion estimation. The paper presents an improved algorithm with the initialized searching point prediction, half-stop principle, beforehand skip mode decision, reduced search window and optional search model. This algorithm for the integer-pixel motion estimation is provided to reduce computation complexity in the block-matching process. Experimental results show that the proposed algorithm can reduce about 64%-73% of computation complexity compared to full search but only with little PSNR degradation.
Keywords :
computational complexity; data compression; decision theory; motion compensation; motion estimation; real-time systems; search problems; video coding; AVS; audio video coding standard; beforehand skip mode decision; compensation algorithm; computation complexity; encoder; half-stop principle; initialized searching point prediction; inter-prediction integer-pixel motion estimation algorithm; optional search model; real-time video codec technology; reduced search window; video compression; Degradation; Educational institutions; Encoding; Entropy coding; Filters; Motion compensation; Motion estimation; Quantization; Video coding; Video compression; AVS; block matching; inter-prediction; motion estimation; motion vector;
Conference_Titel :
Broadband Network & Multimedia Technology, 2009. IC-BNMT '09. 2nd IEEE International Conference on
Conference_Location :
Beijing
Print_ISBN :
978-1-4244-4590-5
Electronic_ISBN :
978-1-4244-4591-2
DOI :
10.1109/ICBNMT.2009.5348472