Exploration of algorithms and implementations for efficient 3D-HEVC depth map encoding

• Autor(a) principal: Sanchez, Gustavo Freitas
• Data de Publicação: 2019
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações da PUC_RS
• Texto Completo: http://tede2.pucrs.br/tede2/handle/tede/8651
• topic: 3D-HEVC; Depth Maps Coding; Intra-Frame Prediction; Inter-Frame Prediction; Encoding Effort Reduction; 3D-HEVC; Codificação de Mapas de Profundidade; Predição Intra-Quadro; Predição Inter-Quadro; Minimização do Esforço Computacional; CIENCIA DA COMPUTACAO::TEORIA DA COMPUTACAO

3D High Efficiency Video Coding (3D-HEVC) standard was the pioneer standard to use depth map coding along with texture coding to achieve high compression ratio and qual- ity in 3D videos. Depth map coding introduced new tools requiring more computational effort compared to texture coding, implying new challenges to encode 3D videos efficiently, espe- cially when the encoding is performed on embedded systems or requiring real-time process- ing. This Thesis proposes algorithms and implementations to improve the depth map coding in the 3D-HEVC standard. The main contributions are divided into three parts. (i) Reduc- tion of intra-frame coding effort, where four new algorithms were designed based on a deep analysis of the time distribution and modes used in the coding process. (ii) Enhancement of bipartition modes, exploring (a) the encoding effort control of the bipartition modes, (b) the parallelism for Intra_Wedge mode, (c) the compression of patterns for wedgelet storage reduction, and d) the hardware design for the bipartition modes decoding. (iii) Reduction of inter-frame coding effort, where three new algorithms were designed to speed up the coding of P- and B-frames. The set of algorithms and techniques developed can be applied indi- vidually or combined to improve the use of computational resources when encoding depth maps.