Research On Video Compression Method Of Unmanned Aerial Vehicle Based On H.265

As the UAV have been widely used in various fields,the quality of aerial video is getting more and more attention.Nowadays,UAV users are pursuing high definition and clarity video,it concomitantly increases the amount of data.However,airborne data link transmission capacity is limited,can not meet the task of transmission such large amount of data.The current compression scheme based on H.264 is difficult to further reduce the amount of data.Instead,the new H.265 standard has a siginificant improvement in performance that can reduce the amount of data by 50%,but also brings a surge in the complexity of operation,the time spend to improve 2~10 times,it is difficult to apply in practical use.In view of this,with combination of characteristics of the UAV aerial video platform,this thesis presents a modified H.265 video compression standard algorithm,which can significantly reduce the time for data compression without changing compression rate.More specifically,it mainly optimizes intra prediction and inter prediction,which are considered as two important processes in video coding process.In the intra prediction algorithm,the main theory of H.265 standard in intra prediction is introduced,that is size-flexible encoding unit which is newly introduced in H.265.Then,it is analyzed that the main reason of leading the complexity of the encode process is the judgement for the final size of CU,and the new algorithm is put forward which use correlation encoding unit of video image and texture feature to narrow the scope of judgement so as to simplify the operation.Compared with the ergodic method in the standard model,the new algorithm reduce the scope of judgement,which make amount of computation reduced.The experimental results show that the algorithm can reduce the time by about 36% on the basis of almost no change in the compression rate.In inter frame prediction algorithm,first the motion estimation process of the video encoding process is analysed,in which the process of motion vector searching spend a lot of time.While the UAV aerial video has some particularity which the whole picture have the same move rule,global motion estimated can be introduced to simplify the calculation,so a algorithm based on six parameter affine model is designed.It is proved by experiments that the algorithm can ensure that the compression rate is not greatly changed,and the complexity of the algorithm is reduced by about 80%.In summary,these algorithms makes full use of the characteristics of the video image and the characteristics of the UAV aerial video.Compared to the standard model of H.265,it reduces time cost in compress process without changing in compress rate when used in UAV platform video compression scheme.