Abstract: A multi-view video coding method includes the following steps: an un-coded group of pictures (GOP) is obtained from an un-coded multi-view video. Wherein, the un-coded GOP includes picture sets of several un-coded views. Several present coding views are selected from the un-coded views according to view levels of the un-coded views. Wherein, picture sets of the present coding views includes several present coding pictures. A target coding picture is selected from the present coding pictures according to view levels of and temporal levels of the present coding pictures. At least a coded reference picture, which is referenced by the target coding picture for prediction, is obtained. The target coding picture is coded according to the coded reference picture. In addition, a multi-view video decoding method is also disclosed.

Abstract: A method for motion search between video frames with multiple blocks is disclosed. The present invention provides an effective edge detection-prediction hexagon search algorithm (EDPHS algorithm). The method uses edge values and the distribution characteristics of motion vector and further overcomes the drawback of the conventional hexagon-based search algorithm (HEXBS algorithm). In this way, the present invention has less search points during searching. Through experiments, the present invention, without affecting the compressed video quality, reduces the number of search points by about 80%˜98% in comparison with the diamond search algorithm (DS algorithm) and reduces the number of search points by about 75%˜97% in comparison with the HEXBS algorithm. The advantages make the present invention suitable for the realizing real-time image compression.

Abstract: A multi-view video coding method includes the following steps: an un-coded group of pictures (GOP) is obtained from an un-coded multi-view video. Wherein, the un-coded GOP includes picture sets of several un-coded views. Several present coding views are selected from the un-coded views according to view levels of the un-coded views. Wherein, picture sets of the present coding views includes several present coding pictures. A target coding picture is selected from the present coding pictures according to view levels of and temporal levels of the present coding pictures. At least a coded reference picture, which is referenced by the target coding picture for prediction, is obtained. The target coding picture is coded according to the coded reference picture. In addition, a multi-view video decoding method is also disclosed.

Abstract: A method for motion search between video frames with multiple blocks is disclosed. The present invention provides an effective edge detection-prediction hexagon search algorithm (EDPHS algorithm). The method uses edge values and the distribution characteristics of motion vector and further overcomes the drawback of the conventional hexagon-based search algorithm (HEXBS algorithm). In this way, the present invention has less search points during searching. Through experiments, the present invention, without affecting the compressed video quality, reduces the number of search points by about 80%˜98% in comparison with the diamond search algorithm (DS algorithm) and reduces the number of search points by about 75%˜97% in comparison with the HEXBS algorithm. The advantages make the present invention suitable for the realizing real-time image compression.