Research On Method Of Aerial Image Mosaic In Low-altitude Urban Areas

Because of the existence of large parallax area,there may be some problems such as local distortion,ghosting or non orthophoto of mosaic results when the low-altitude aerial images are stitched based on the traditional image mosaic method.The existing methods can only eliminate or reduce the problem of ghosting and misalignment caused by image parallax to a certain extent,but they do not consider the preservation of orthographic projection of mosaic results.In order to eliminate the bad influence of the large parallax area in the image mosaic and generate an approximate orthophoto panorama of the low-altitude aerial image,this paper takes the low-altitude urban aerial image as the background and proposes the Large-Parallax Patch.The core idea is to treat the large parallax area and the low ground object area separately,so as to avoid the interference of large parallax on image mosaic;at the same time,the large parallax area is screened to ensure that it has the best orthorectity in the mosaic result.Therefore,this paper first proposes a large parallax area positioning algorithm based on feature matching: using the VFC algorithm to distinguish correct and incorrect matching points,and then according to the matching and distribution of feature points in the large parallax area,using the grid method to determine the large parallax area grid through the grid screening,and finally using the region growth to obtain the complete location of the large parallax area.Then the other areas except the large parallax area are unified as low ground object areas for mosaic.At the same time,the image with the best orthophotos is screened from the sequence images of the large parallax area,and patched into the stitching result image of the low ground object area,and the final mosaic result is obtained.When no DEM data is needed,make full use of the information of the original aerial image itself to generate a panorama similar to orthophoto.In order to enhance the flexibility and generality of the algorithm,this paper designs a mosaic system with adjustable parameters and interaction,so that the algorithm can be better applied to different types of image mosaic.Experimental results show that compared with the orthophoto generated by Pix4 D,the proposed method has better image quality and lower time consumption;compared with the panoramic mosaic generated by PTGUI,the large parallax area has no ghosting and misalignment,the mosaic effect is better;compared with the APAP algorithm,the speed is faster,the large parallax area has better ortho,which obviously reduces the occlusion problem of surrounding buildings by tall buildings.This will facilitate more accurate flow statistics,emergency feedback,and urban traffic monitoring.