Fast Depth Intra Coding Based on Decision Tree in 3D-HEVC

The coding units (CU) partitioning in the 3D extension of the high efficiency video coding standard (3D-HEVC) is recursively conducted on different block sizes from <inline-formula> <tex-math notation="LaTeX">$64\times 64$ </tex-math></inline-formula> to <inline-formula> <tex-math notation="LaTeX">$8\times 8$ </tex-math></inline-formula>. Besides, the depth coding in 3D-HEVC introduces several new coding tools for each CU to improve the coding efficiency, however, with great computational complexity. It is noted that only a small number of the CUs in recursive partitioning are encoded into the final bitstream. Among these CUs, the CUs with Intra <inline-formula> <tex-math notation="LaTeX">$2\text{N}\times2\text{N}$ </tex-math></inline-formula> or Intra N <inline-formula> <tex-math notation="LaTeX">$\times $ </tex-math></inline-formula> N as optimal modes have a very small proportion. In this paper, we thus propose an early determination of depth intra coding, where the coding stage of Intra <inline-formula> <tex-math notation="LaTeX">$2\text{N}\times2\text{N}$ </tex-math></inline-formula>, Intra N <inline-formula> <tex-math notation="LaTeX">$\times $ </tex-math></inline-formula> N or CU Splitting for the CUs could be early skipped based on several Decision Trees. Simulation results show that, with restrict the results from the tree leaves by different Gini thresholds, the proposed algorithm could save 41.14%–71.63% of the depth coding time with only a slight increase in BDBR.