Research On Synchronous Stitching And Change Finding Based On UAV Image

The most stringent land management system is performed in our country and it is very important to early discover and timely dispose the illegal land.Aiming at the problem of occupation,random approval and serious abuse of land,this paper is dedicated to the research of synchronous stitching and change finding based on UAV image,which can realize the rapid automatic discovery of suspicious objects and the dynamic monitoring of the condition of the land.The main contributions are as follows:(1)The preprocessing problem of large-scale old time phase images is studied.The massive image data of the large-scale old time phase information is extracted and stored in the library through the establishment of library which can finish the retrieval and organization of large-scale old time phase image feature points and ensure rapid matching positioning between the new and old time phase images.(2)A coarse-fine double matching recursive model is proposed,which is based on geographic information provided by the UAV POS system and the feature point information of new and old time phase images.It can realize rapid matching positioning and guarantee for the real-time fast find suspicious object.(3)A detection algorithm of changed land combined with the color,texture and feature points is proposed.It can realize the rapid automatic discovery of suspicious objects which does not need to splice the UAV images by multi-angle comprehensively comparing and detecting the new and old time phase images based on the color,texture and feature points of the image information.(4)A synchronous stitching algorithm based on old time phase image is studied.The new time phase image can be achieved which does not need to lay a large number of ground control points by synchronized stitching changed UAV image based on the old time phase.It can update the change of information of the old time phase image and provide more effective information reproduction for the subsequent land patrol.