Non-orthogonal transform coding method for video signal

2016 Fiscal Year Final Research Report

• Project/Area Number: 26420342
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Communication/Network engineering
• Research Institution: Kanagawa Institute of Technology
• Principal Investigator: Hatori Yoshinori 神奈川工科大学, 情報学部, 客員教授 (20361769)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: 映像信号符号化 / 非直交返還符号化 / 非因果的符号化 / ハイブリッド符号化 / 視点内挿

Outline of Final Research Achievements

In this research, we focus on optimization of the coding method which expresses the two major element technologies used in video coding, predictive coding and orthogonal transformation processing as a function of "non-orthogonal transformation" in a unified manner I tried it. In addition to presenting the design possibility and theoretical advantage of non-causal coding, we aimed to realize the coding performance higher than the currently used HEVC method. By realizing extension of the non-causal coding method to inter-frame motion compensated predictive coding (B picture coding processing), depending on the picture, from 1.77 dB to 0.13 dB from the existing HEVC method with PSNR value It is shown that an excellent encoding method can be realized. Furthermore, we investigated non - orthogonal transform coding method suitable for interpolation between viewpoints of multi - view video.

Free Research Field

通信・ネットワーク工学