Project 5A: IMAGE WARPING and MOSAICING
1. Shoot and digitize pictures
All pictures are taken with my smartphone. The pictures will be
shown in the corresponding sections where they were used.
2. Recover homographies
The homography was solved by solving the linear equations below.
The a,b,c,d,e,f,g, and h are unknowns. The constant i is set to 1. We need
at least four points to solve for the eight unknowns. We obtained
more than four points to have an overdetermined system and solved it using
linear least square.
So
3. Image Rectification
Below we showed images before and after rectification.
Image One |
|
Image One Rectified |
|
Image Two |
|
Image One Rectified |
|
4. Image Mosaic
Below we showed the original images and image mosaic
Image Left |
Image Center1 |
Image Center2 |
Image Right |
|
|
|
|
Mosaic |
|
Image Left |
Image Center1 |
Image Center2 |
Image Right |
|
|
|
|
Mosaic |
|
Image Left |
Image Center1 |
Image Center2 |
Image Right |
|
|
|
|
Mosaic |
|
4. What I learned
Very interesting project. This helps me to practically apply what've learned
about homography to my own dataset and implemented the corresponding functions
basically from scratch.