Title :
Saliency based perceptual HEVC
Author :
Yiming Li ; Weihang Liao ; Junming Huang ; Da He ; Zhenzhong Chen
Author_Institution :
Wuhan Univ., Wuhan, China
Abstract :
Saliency represents the probability of human attention over the image therefore is important in understanding the importance of different areas in the image. In this paper, we consider how to utilize the saliency information in HEVC video coding. The graph-based visual saliency is incorporated with the quantization control in HEVC to reduce the bit rate of compressed video by using larger quantization parameter for the coding unit which has lower probability of attention. As shown in our experiments, the proposed method achieves up to 12% bit rate reduction without perceptual quality loss.
Keywords :
graph theory; quantisation (signal); video coding; HEVC video coding; graph based visual saliency; human attention; perceptual quality; quantization control; quantization parameter; saliency based perceptual HEVC; saliency information; Bit rate; Circuits and systems; Encoding; Quantization (signal); Transform coding; Video coding; Visualization; HEVC; quantization parameter; saliency;
Conference_Titel :
Multimedia and Expo Workshops (ICMEW), 2014 IEEE International Conference on
Conference_Location :
Chengdu
DOI :
10.1109/ICMEW.2014.6890644