Research On Image Mosaic Algorithm Based On UAV

UAV technology is a hot project in recent years,and drone vision research is the focus of UAV research.According to the characteristics of the image acquired by the drone and the limited shooting space of the image,it has received more and more attention in the field of application and scientific research.This paper starts with the research on the image mosaic technology of UAVs,compares the research status of UAV image mosaic technology at home and abroad,and determines the main purpose and research methods of this paper.The content and innovation of the thesis research are as follows:Firstly,the image processing algorithm and the splicing basic algorithm of the drone are studied.In this paper,the processing techniques of UAV images,such as histogram matching algorithm,image neighborhood filtering and image enhancement transform method,are studied.The basic theory of image splicing is studied.Through image splicing characteristics and process analysis,image splicing is studied.The process,from image preprocessing,feature point extraction,to image registration,image fusion,researches various algorithms in the process,converts multiple UAV images into the same coordinate system,and uses image fusion to form a whole image.Secondly,this paper focuses on the process and improvement of the image mosaic algorithm of drones.First,the feature points that need to be extracted are obtained by a general Harris corner detection algorithm.Second,the SURF(Speeded Up Robust Features)algorithm is used to improve the scale invariance of the Harris corner detection algorithm.Thirdly,aiming at the performance defect of the SURF algorithm in efficiency,a statistical algorithm for historical feature points is proposed to improve the SURF algorithm and achieve high-speed extraction of feature points.After the feature point extraction,coordinate correspondence between the processed image and the reference image is obtained by using coordinate affine transformation and RANSAC(Random Sample Consensus)algorithm.Fourth,the image is blended using color conversion and color overlay.Finally,an algorithm experiment was carried out.The purpose of the algorithm and the various steps of the algorithm implementation,the experimental results are obtained through these algorithm steps and the success or failure of the algorithm is analyzed from the correctness and efficiency.