The Embedded System For Intelligent Video Monitoring In UAV

Unmanned aerial vehicle(UAV)is a kind of unmanned air vehicle which is controlled bywireless remote devices or programs. Compared with the gerneral plane, UAV has manycharacteristics such as low cost, high efficiency, so it is widely used in industrial produce andmilitary defense. The purpose of the video surveillance system in UAV are receiving videoSequences, store the image datas, rea-time processing, and so on. The performance of thissystom decides whether the uav can be applied effectively. When the UAV executes the videomonitor task, the dithering and fuzzy phenomenon will be contained in the acquired images forthe external environment and their own factors. As a result, the following works can notcomplete successfullyOn basis of actual requirements, key technologies of the HD video processing platform arestudied in this paper. In the algorithm of electronic image stabilization, an improved method ofimage registration is presented. Feature points are detected and matched by setting searchwindow scope and correlation coefficient threshold. On basis of the feature point matching,motion estimation and compensation are then performed, and video images stabilized can be got.In the algorithm of image mosaic, feature points are detected, and then they matched by NCCmethod. The outliers are removed based on the statistical regularities of matching points. Theimproved RANSAC for feature points exaction can greatly support the real time of thealgorithm. HD video processing platform is designed based on DSP and ARM which can worktogether by using Davinci technology.HD video processing platform designed in this paper can run the algorithms which arementioned above. The experimental result shows that the improved algorithm of electronicimage stabilization and the improved RANSAC algorithm of image mosaic are fast andeffective. HD video processing platform can be future used in the field of earth observation.