Optimization of Parallax Image Projection Based on SIFT
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...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Harbin University of Science and Technology Publications
2020-08-01
|
| Series: | Journal of Harbin University of Science and Technology |
| Subjects: | |
| Online Access: | https://hlgxb.hrbust.edu.cn/#/digest?ArticleID=1852 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | 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 |
|---|---|
| ISSN: | 1007-2683 |