Frame-compatible format, packing two neighbouring views into one frame, is considered as a promising solution for 3D distribution in the existing system. In this paper, a fast Intra encoder is designed for frame-compatible format coding based on the content similarity and proposed to reduce the computation complexity. With a relative shift, statistics analysis proves the high prediction correlation between the two packed views and qualifies the first coded view provide prediction mode as the reference for the second view predic-tion. The proposed scheme enables the prediction of the sec-ond view's MB only perform candidate Intra modes and candidate directions of each, according to the reference MB in the first view. An average of 75.69% complexity reduction of encoding the second view can be achieved with compara-ble compression efficiency. With no requirement to change the decoder of existing system and there is little effort for this encouraging format, this proposed algorithm improves the efficiency of frame-compatible format coding, and con-tributes to the delivery of 3D service.
|出版ステータス||Published - 2011|
ASJC Scopus subject areas