UAV Video Compression Algorithm Based On Hierarchical Description

Because of UAVs' outstanding performance in mobility and strong vitality with low cost, they are widely used in military and civil fields. Videos are considered as the most efficient and intuitive carriers of information. They play an irreplace able role in a variety of tasks. However, the transmission of huge amount of UAVs' high definition video data with limited bandwidth has been a bottleneck for the application of UAVs, while the current standard compression algorithms such as H.264, MPEG-2 etc. cannot guarantee the quality of transmission and overcome the above difficulties simultaneously. Therefore, it is very important to study the video compression algorithm for UAV applications. In this paper a novel UAV video compression algorithm based on hierarchical description is proposed to solve the problem, which is able to provide a more flexible compression and description method to adapt more complex application request.The research of this paper can be summarized as the following aspects:(1) Considering the plane background and consistent motion characteristics of the UAV videos, a moving objects segmentation method based on global motion estimation is proposed, which is able to separate moving objects rapidly and precisely.(2) Based on the analysis and mining of the UAV videos' features, a hierarchical video description based on moving objects segmentation is proposed, which is consisted of the background layer, the object layer and the enhancement layer, and each layer can be independently encoded. The background's displacement is mainly caused by the airborne camera. Therefore, to compress the background layer with high rate and low complexity, this paper uses global motition estimation based on six-parameter affine model. The object layer is compressed by modified H.264. The enhancement layer is composed of residual of the first two layers, which usually uses small quantization parameters to guarantee high quality of the video.This paper uses simulation videos and actual aerial videos to verify validity and real-time of the proposed algorithm respectively. The results are compared with H.264. The experiments demonstrate that compared with H.264, the proposed algorithm can provide more effective video transmission of higher quality with low bitrate and similar subjective quality with high bitrate. In addition, the advantage of the proposed algorithm is its more flexible bitrate adaptability.