Panoramic image stitching
A panoramic image can be formed by stitching together two images with overlapping scenes. The overlapping images taken usually have common points called interest points which can be extracted to stitch the two images together. The interest points can then be matched and merged to form the panoramic...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
2016
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/68315 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | A panoramic image can be formed by stitching together two images with overlapping scenes. The overlapping images taken usually have common points called interest points which can be extracted to stitch the two images together. The interest points can then be matched and merged to form the panoramic image. [1]
This report will describe how the interest points are located and extracted using Harris Corner Detector with the help of a 2-by-2 matrix and using eigenvalues to determine whether the points are suitable enough. Cross correlation will then be used to match the interest points from the two images together. RANSAC will then get rid of the outliers leaving only the inliers to blend the two images together.
This report will discuss on how different scenarios will affect the resulted stitched image. Images that are rotated, scaled and also those with their illumination and perspective changed will be tested. |
|---|