| Stereo video is popular due to its realism. The principle of binocular parallax of human eyes is utilized in stereo video technique. Left eye views the left image and right eye views the right images. So the binocular parallax was achieved. The brain can gain the depth information of the image. In the end, people can enjoy a three-dimensional visual image. The application of stereo vision is more and more extensive, its application includes 3DTV, 3D video conference, remote medical service and so on. The amount of data increase obviously, it is difficult to store and transmit enormous of video data, especially in high resolution and real-time system. Therefore, in the future, one of the keys for utility of stereo video is how to compressing stereo video data efficiently and reducing the amount of data. The stereo video not only exists the intra-frame spatial redundancy and inter-frame temporal redundancy, but also exists inter-sequence redundancy. It needs to exploits the inter-sequence redundancy.H.264 has better coding efficiency and better adaptively to net transmission, it can adapt to different network to deliver stereo video. Since the basic systems of H.264 can adapt to the use of IP and network, it is important for the data transmission of multimedia information through internet or WAN information through mobile network. The key techniques of H.264 are intra-prediction,inter-prediction,integer DCT transform and quantification,CAVLC and filter etc. For making use of the high performance of H.264, in this paper, we propose a new stereo video coding method based on H.264. The algorithm can keep detail in the image and reduce the coding bitrates at the same time, this paper analyzes the algorithm option in intra-frames or inter-frames coding of H.264. Nowadays, stereo video compression has been paid attention by many researchers. ATTEST proposed a novel stereo video communication that joint transmission of monoscopic color video and associated per-pixel depth information. 3D scene can synthesized in real-time at the receiver side. The research of depth image-based synthesis includes the estimation of depth image and view synthesis algorithm.Since depth video is not affected by illumination, the texture, color and shadow information do not exist. These characteristics may produce high coding efficiency in depth video coding. The depth video has something in common with the color video. The boundary of the depth video has similar shape with boundary of the color video. Based on the correlation, we proposed the joint coding method of video plus depth based on H.264. The depth video takes the YUV format 4:0:0.Because most parts of depth map are flat, most intra-prediction modes of depth's macroblocks are chosen 16×16 mode. For I frame of depth video, we consider the 16×16 mode and add a flag to bitstream in order to reduce the bits. When the prediction modes of color video and depth video meet the proposed algorithm, the flag is 1, otherwise flag is 0. When the flag equals 1, prediction modes of depth don't need to deliver, it can be get in the decoder by making reference to color video. When the flag equals 0, prediction modes of depth need to deliver. For p frame of depth video, we consider intra-prediction and inter-prediction respectively, then consider them together. When we only consider inter-prediction, the macroblock of color video and depth video is same, the flag is 1, the prediction mode and motion vector of depth do not need to write in bitstream. The decoder reference the color video to get the prediction mode and motion vector. Otherwise the flag is 0, the prediction mode and motion vector of depth video need to write in bitstream.We consider the intra-prediction and inter-prediction together, add two flags to bitstream of every depth macroblock. When the intra-prediction mode of depth is 16×16 mode, the first flag is 0,the second flag is 1, img->i16offset which doesn't need to add 6 is encoded into bitstream. When the prediction mode of depth is inter-prediction mode, its prediction mode is same as color video's prediction mode. The first flag is 1, the second flag is 0, the prediction mode and motion vector of depth do not need to write in bitstream. When bitstream of depth do not satisfy these two conditions, two flags all equal 0.All experiments are carried out under VC++6.0, the proposed method considers a block as a unit to perform prediction, transform, quantization and entropy coding to get rid of various correlations in the stereoscopic video, we compared the coding efficiency of H.264 with the proposed method. The experimental results show that the proposed method can decrease the encoding bitrate effectively and keep detail in the video. |
