CS 194-26: Image Manipulation and Computational Photography

Project 6A: Image Warping and Mosiacing

Ryan El Khoury, CS196-26-aah

Overview

Coming soon...

Shooting the Pictures & Correspondence Points

Explain how correspondence points are chosen

Recover Homographies

Explain what a homography is and how we can "recover" it

Warp the Images

Explain what it means to warp an image using a homography, forward vs. inverse warping and which one you choose + why

Image Rectification

Explain what it means to rectify an image, and how (on a high level) that happens.

Chess board, seen from a slanted angle.
Correspondence points on chess board and intended rectification shape.
Rectified chess board.
Window/TV, seen from a slanted angle.
Correspondence points on room scene and intended rectification shape.
Rectified window/tv.

Mosaicing

These pictures were taken in my apt and the surrounding vicinity. The images were stitched together using Laplacian blending. The brightness of the overlapping portions of the image were probably due to imprecise averages between the two images.

Left image of interior.
Right image of interior.
Left image warped to right image plane.
Stitched interior.
Left image of balcony.
Right image of balcony.
Left image warped to right image plane.
Correspondence points picked for balcony images.
Stitched balcony.
Left image of stairs.
Right image of stairs.
Left image warped to right image plane.
Intermediate level of Laplacian blend.
Stitched stairs.

Summary

The most interesting thing I learned from this project was how much the information contained within an image can be shifted to give an entirely new perspective. In particular, rectifying part of an image to be "straight" not only straightens out the specified area, but also the entire image. On top of that, the result does not even look that warped/stretched!