Tencent Technology (shenzhen) Company Limited

Chenchen Yu

Wen Gao

Ronggang Wang

Jing Lv

Jingchang Chen

Jie Wan

Quanzhan Zheng

The present application provides a video encoding method that includes setting frame types for a video sequence; obtaining a B frame; determining whether a current macroblock of the B frame satisfies a Direct prediction mode, and if yes determining whether the current macroblock satisfies a Skip prediction mode; if the current macroblock does not meet either mode, computing at least one of a mode cost after performing motion compensation on the current macroblock using two bidirectional prediction motion vectors obtained in the Direct prediction mode; a mode cost after performing motion compensation on the current macroblock using a forward prediction motion vector obtained in the Direct prediction mode; and a mode cost after performing motion compensation on the current macroblock using a backward prediction motion vector obtained in the Direct prediction mode; and selecting a mode with a smallest cost as an optimal prediction direction to encode the current macroblock.

Growing community of inventors

Shenzhen, China

Chenchen Yu