Researchs On HEVC Based Video Coding Mechanisms And Video Transmission Mechanisms Over Wireless Networks

With the rapid development of social networks and mobile devices, the transmission proportion of video over the networks is increasing. High definition and ultra high definition video technology and their applications are influencing the views and ways of our life and work. At the same time, the High Efficiency Video Coding(HEVC) as a new generation video coding standard promotes the development of video. Due to the character of direct, specific, real-time access to information, it is convenient to apply video in our life and work. Along with the development of mobile phone application software, video sharing is becoming a way for people to express emotions. Moreover, the development and convenience of communications meet our needs for the network at anytime and anywhere. Not only the single priority but also multi-priority services are supported by the networks now, which benefits the transmission for multi-media data.The HEVC based coding mechanisms and transmission mechanisms over wireless network for the video are studied in this dissertation. The major contributions of this dissertation are listed as follows:(1) In order to reduce the high computational complexity in HEVC, the the methods of machine learning are introduced. And a coding mechanism based on decision tree is poposed for HEVC. The Coding Unit(CU) depth splitting problem is viewed as a classification problem in this mechanism. The spatial correlations in the video frames are utilized. The splitting information of neighbor CU is extracted for the off-line learning in decision tree model. Then the decision tree based categorizer is obtained which can be used to judge the splitting of CU. If the classification condition is met, the computation of rate-distortion(RD) cost is skipped. Therefore, the computation complexity in HEVC is reduced. Experimental results show that the proposed mechanism achieves reduction of the computational complexity in an average of 32 percents, by comparing with the original HEVC, while still maintaining high video quality(2) In order to reduce the high computational complexity in HEVC and avoid an additional off-line time to study the rules in the the methods based on machine learning, the statistics is introduced, and another real-time coding mechanism based on the adaptive CU depth range is proposed for HEVC. The CU splitting procedure and the distribution of CU depth are studied in this dissertation. Then we difine the adaptive CU depth range. The encoders can only calculate the RD cost at the depths which are included in the specific range. RD cost calculations at the CU depths outside of the specific range can be skipped. Experimental results show that the proposed mechanism achieves reduction of the computational complexity in an average of 16 percents, by comparing with the original HEVC, while still maintaining high video quality.(3) Since the collisions between different kinds of services transmitted over the wireless network reduce the network throughput, the concept of hybrid slot is introduced. And a video transmission mechanism of hybrid slot is proposed. A concept of hybrid slot is defined which contains several slots with diffrent priorities. We difine the hybrid slot as the basic transmission slot. Services with different priorities are allotted to slots with deifferent priorities. The collisions between services with different priorities in a hybrid slot are avoided. So it can reduce the collisions and retransmissions among each service and improve the throughput of network. The mechanism of hybrid slot video transmission promotes the network throughput in an average 15 percents.