Research And Implementation Of Feature-based Remote Sensing Image Stitching Technology For UAVs

UAV remote sensing image stitching technology plays an important role in disaster warning,emergency response and other fields.In recent years,it has become a research hotspot.This thesis studies the feature-based stitching technology of UAV remote sensing images,and the main work completed are:Firstly,an algorithm for stitching remote sensing images of unmanned aerial vehicles based on improved SIFT features is proposed,which first performs rough matching of SIFT features on the images,then introduces matching quality ranking at the fine matching features of the RANSAC algorithm,and preferentially the selection of matching points with good quality for matching,so as to effectively remove the wrong matching point pairs,and finally use the projection-similar transformation model to complete the model transformation to complete the stitching.The algorithm effectively improves the distortion,blur and inaccurate alignment in the stitching result,and the stitching speed is also effectively improved.Secondly,for the problem of high time-consuming stitching,it is proposed to use the ORB feature-based UAV remote sensing image stitching algorithm for stitching,which first extracts and describes the ORB feature,then uses the K nearest neighbor method for matching,and finally,the projection transformation model is used to complete the stitching.Compared with SIFT algorithm,the speed of stitching is greatly improved.Finally,a set of UAV remote sensing image stitching system is designed and implemented.The system is equipped with three modules: image reading,image preprocessing and image stitching.After the image is read in,it can be preprocessed first,and then the stitching scheme can be selected for stitching.