Attention region based rate control algorithm for 3DVC depth map coding
Pei‐Jun Lee, You-Zhi Xiao, Shao-Ru Yao · 2016
This paper proposes an algorithm that can adjust the essential quality for different ROI and further control the bit rate. We select three characteristics for classification, namely foreground, the edges of objects, and dynamic region as the interesting area of eyes. In the region of high interest, we will give a smaller quantization parameter to achieve a better image quality; on the contrary, in the region of low interest, we then give a larger quantization parameter to reduce the amount of data. According to our last experimental results, our scenes would been correctly classified out, which was decreased by 10% and 5% of the bit-rate in comparison with the bit-rates of 3DVC reference software ATM6.0 and the references in bit-rate control.