A Study Of 3D Video Compression And Transmission Based On H.264

With the improvement of people’s living standards and the popularization of the science and technology, 3D technology has increasingly approach our daily life. Especially with the arrival of the trend of 3D movie and television, people are more and more used to enjoying the visual impact and immersive senses of reality brought by the 3D display, and hope the 3D technology extend to other fields, such as video conference, telemedicine, etc. Besides, because the basic principle of the traditional 3D video is providing two signals respectively simulating the left and right eye to achieve the 3D effect, it can’t truly reflect the real 3D scene. Therefore, people hope to describe the actual scene more truly through acquiring the depth information.The development of consumer depth devices, especially the depth acquisition devices led by the Microsoft Kinect, provide the possibility of practical use for such applications, which can acquire the traditional color video as well as the depth information of the various articles in the scene by adopting the infrared camera and optical modulation technology so as to provide more realistic 3D scene experience for the users. The purpose of this paper is to design a color+depth video compression and transmission system that can be realized by the project.The 3D stereo video system will bring a large amount of data when loading depth information, which will bring new technology challenges to the compression encoding and network real-time transmission of the video. Firstly, to avoid the “surface effect” of the depth information due to lack of accurate reading, the depth information usually adapts the 16-bit dynamic range. While the current mainstream coding standards do not support this kind of high-dynamic video coding and there is no relevant international coding standard so far. Therefore, the first question to be solved in this paper is how to achieve the high efficient compression encoding by using the existing H.264 standard through remapping of the depth information. By exploring the depth video coding, this paper comes up with several mapping schemes and gets a better depth coding method that can be realized by the project through large amount of statistical experiment.Furthermore, the increase of the data brings forward higher requirement for the network transmission, especially the real-time and data security. To make sure the real-time of data transmission, we set the cache mechanism both in client and server side, which can make sure the fluency of the video; to avoid the package loss caused by the limitation of network bandwidth, we set the feedback mechanism to make sure the stability of the data during transmission. We unpack and pack the coded data, put it in the cache of the server side, and sent it out from the cache when receiving the send order; save the RTP received from the network into the cache of the client side for decoding after unpacking.