Research On Remote Sensing Image Mosaic Technology Of UAV For Bridge Detection

UAV remote sensing image mosaic technology is based on UAV as a remote sensing platform.The sensor acquired sequential images are processed by computer,and two or more images are mosaic to form a large perspective image.This technology solves the problem that a single remote sensing image can not cover the target area and can not meet the actual operation needs.Because of its high efficiency,low cost and reliable mosaic image quality,it has been widely used in engineering surveying,traffic monitoring,military investigation,disaster investigation and other fields.In this paper,the crack image collected by UAV in bridge detection is taken as the research object.Aiming at the two key technologies of image registration and image fusion involved in the process of image mosaic,an in-depth study is carried out.The main research contents are as follows:(1)The method of image registration based on point feature is studied.In the detection of feature points in adjacent images,according to the characteristics of bridge crack images collected by UAV,an improved feature point detection algorithm is proposed.This algorithm combines SIFT algorithm with SURF algorithm,uses the method of SURF algorithm to construct scale space and locate feature points,and uses SIFT algorithm to generate feature directions and feature descriptors to meet the characteristics of bridge crack images.Feature extraction.The improved algorithm improves the calculation speed and the matching accuracy between images.In image feature matching,Euclidean distance ratio method is used to measure similarity,and k-d tree data structure and BBF algorithm are used to achieve rough feature matching.After rough matching,there are mismatched point pairs in the image.RanSAC algorithm is used to eliminate the outliers and to purify the feature points.The results of image matching show that the algorithm used in this paper achieves precise matching of feature points.(2)Pixel-level image fusion is studied.Based on the analysis of non-multi-resolution and multi-resolution image fusion methods,a Laplacian fusion method based on the best stitching seam is proposed.The algorithm first searches for the best stitching line,and then uses the Laplacian method of multi-resolution analysis for image fusion,which solves the problem of blurring and stitching seams in overlapping areas after image fusion.The experimental results show that the fusion effect of the improved algorithm is better than that of other algorithms.