中文摘要：

针对高空无人机航拍序列图像的实时图像压缩问题,设计了一种基于单应变换的可精确定比的高压缩性能低计算复杂度航拍序列图像压缩算法.该算法使用机载惯导数据进行编码帧组划分并获得中间帧相对于首尾帧的初始单应变换,进而通过多子区匹配实现对单应矩阵的快速精确求解,然后通过单应变换来实现对中间帧的整体运动估计和补偿,最后用JPEG2000算法对帧组数据进行整体压缩.实验结果表明,相对于传统算法,该算法具有计算复杂度低、压缩比高、保真度好、可精确定比压缩的优点.

英文摘要：

Focus on the image compressing problem of unmanned aerial vehicle,a novel image compressing algorithm based on homography transformation was proposed.The image sequences were divided into frame-groups according the data of airborne inertial navigation systems（INS）,and the intermediate frames was bi-directionally predicted by the first-frame and the end-frame with homography transformation.The homography matrix was got approximately by the airborne INS firstly and then was accurately computed by fast multiple template matching.Finally the first frame and the residual data of the intermediate frames in each frame-group were holistically coded by JPEG2000 with global optimization.The experiment results show that the proposed algorithm was with high compression performance,low computational complexity and excellent capacity for code-size control and will has good prospect in engineer.