Project 6A: Image Warping and Mosaicing

Samantha Lai | cs194-26-afr


This is the first part of a larger project about auto-stitching panoramas.

Recovering homographies

I used this matrix equation to solve for the homography between photos.


In this part of the project, I solved for the homography that mapped the corners of a perpective view of a rectangular object to a frontal-parallel view of the object.

Blending Images Into a Mosaic

In this part, I solved for a homography that related two different perspective views to stick together two images in each mosaic with the same center of projection. I blended the images using a "1 level pyramid" by creating a binary mask and smoothing the stark edge of the mask using a Gaussian filter.

Summary of Project 6A

So far, we have implemented a way to warp images with the same center of projection to a common projective plane and stitch together panoramas. This approach requires us to manually define points of correspondence.