Research On Automatic Mosaicing For Lmages Obtained From UAV

UAV（Unmanned Aerial Vehicle）remote sensing technology is more and more widely usedin investigation and evaluation of geologic disasters,because of the advantages ofmobility,timeliness and so on.At present, the image we need to research of the area interested ismosaiced by hands.Due to the immense data faced typically,we can not get the information of thearea hitted by unexpected incidents with the image mosaiced by hands timely.Against thisbackdrop,the paper tried to present an image mosaicing algorithm based on SIFT.The maincontext of this paper could be list as follows:1.The algorithm SIFT was introduced,which realized image features extracting and imagematching. Not only the principle and steps of RANSAC were illuminated,but also A modifiedRANSAC was put on in a bid to get rid of outliers,which make the speed of RANSAC faster andthe result of SIFT more reliable.2. Floyd and disjoint sets were introduced to determine the number of same places. Finally,the result of number of same places based on SIFT have been obtained by experiment.3. The principle of holistic error compensation was illuminated in detail,and the model ofholistic error compensation was deduced,with same places assigned weigh according toamount.The model was solved by LMS.It was show by experiment that strip deformation couldbe avoided by holistic error compensation.4.Based on the researches above,a system of automatic mosaicing has beendeveloped,which has applied in some research projects.