Fast mode selection based on texture segmentation and view prediction in JMVC

In order to reduce the encoding complexity and improve the random access performance of Joint Multi-view video coding (JMVC), this paper proposed a fast mode selection algorithm based on texture segmentation and a new prediction structure. In the fast mode decision algorithm, macroblocks are classified according to their texture complexity, and each frame is divided into flat background region, complex background region and foreground region by an adaptive threshold. Different segmentation methods are used in different regions to reduce the candidates for mode decision. Predictive modes are filtered further according to the texture direction of macroblocks, which can eliminate some unnecessary modes and improve the speed of mode searching. In the proposed prediction structure, the selection of I-view is adjusted and some inter-view prediction is removed. Compared with JMVC, experimental results show that the proposed algorithms can improve random access performance and reduce up to 65% of the computational time, while maintain high RD performance.