This dissertation presents a fast depth map coding for 3D-High Efficiency Video Coding (3D-HEVC) based on static Coding Unit (CU) splitting decision trees. The proposed solution is based on our previous works and avoids the costly Rate-Distortion Optimization (RDO) process for depth maps coding, which evaluates several possibilities of block partitioning and encoding modes for choosing the best one. This coding approach uses data mining and machine learning to extract the correlation among the encoder context attributes and to build the static decision trees. Each decision tree defines if a depth map CU must be split into smaller blocks, considering the encoding context through the evaluation of the CU features and encoder attributes. The results demonstrated that this approach can halve the 3D-HEVC encoder processing time with negligible coding efficiency loss. Besides, the obtained results surpass all related works regarding processing time and coding efficiency. The results reported in this dissertation were published in three journals and two events, besides generate a patent deposit. These products have the master student as the first author.
Improvement of JPEG XL Lossy Image Coding Using Region Adaptive DCT Block Partitioning Structure
