CS 194-26 Spring 2020

Project 5: Image Warping and Mosaicing

Alex Shiyu Liu, cs194-26-afw

Overview

This project explores image warping and mosiac stitching through homography transformations.

Shoot the Pictures

The images I shot myself are below:

Recover Homographies

In this part, I found the 3x3 homography matrix to map one image to the other. I found the equation below from Prof. Collins CSE485 course notes at Penn State. This is an alternate representation of the equation x'=Hx that accounts for additional points. The additional points create an overdetermined system that result in a more stable H matrix once the system of equations is solved. I used either 8 or 4 points for all of the examples below.

Image Rectification

In the example image below, I rectify the image to align the right painting to be facing straight on.

In the example image below, I rectify the image to align the top right square of the door to be facing straight on.

Image Mosiac

In the example image below, I rectify the image to align the right painting to be facing straight on.



Conclusion

I learned a lot about the homography transformation and its applications to image stitching, mosiacs and panoramic pictures.