Abstract:In order to solve the problem of misalignment and ghosting in image stitching due to large parallax

this paper proposes a parallax image projection optimization algorithm based on SIFT First

the feature points of the image are acquired through SIFT

and then the mismatched points are removed from the image by the random sampling consistency algorithm

and the projected matrix calculation is performed on the removed feature points. Finally

the image is pieced together by the fade in and fadeout algorithm Seam elimination operation. In order to improve the accuracy of image stitching

after removing the mismatched points in the image

the algorithm in this paper calculates the overlapping area between the images and calculates the distance after the projection of the boundary points of the area to find the optimal projection matrix. Experimental results show that the algorithm can effectively improve the alignment rate of parallax images and improve the quality of image stitching