Efficient depth coding in 3D video to minimize coding bitrate and complexity

Abstract

In three-dimensional video (3DV) coding, color videos and depth maps both need to be coded. Usually, two-channel coding method is used in the system of 3DV coding which encodes color videos and depth maps based on two parallel codec implements. The complexity and hardware requirements are nearly two times higher than coding the 2D color video. Meanwhile, depth maps in 3DV usually are estimated from color video data and the estimated depth itself can be very noisy. In this paper, we propose fast and effective depth coding to minimize depth coding bitrate and the coding complexity. The 3 × 3 bilateral filter is first utilized to pre-process depth maps to reduce noises from the depth estimation procedure, and thus unnecessary bits to code depth noises can be reduced. Meanwhile, there is a high correlation among motion information from color videos and depth maps. Coding information including motion vectors and the prediction mode is drawn from the color video to accelerate the mode decision procedure of depth coding and reduce the temporal variation of depth maps. Experimental results show that the proposed algorithm can reduce 70% computational complexity of depth coding and 20% depth bitrate.