Intro to Computer Vision and Computational Photography, Fall 2021

Project 4 Part 1: Image Warping and Mosaicing

SOPHIA SONG, CS194-26-ach



Overview

In this project, the goal was to understand and implement image warping and mosaicing using concepts like homographies.

Part 1: Shoot the Pictures

I decided to shoot photos of the Haas School of Business here at Berkeley with the goal of creating a panorama of the campus.

Image 1
Image 2
Image 2
Image 4

Part 2: Recover Homographies

Then, I wrote a function to generate the homography between two images and their corresponding points. For this project, I chose ten points in each of the images to use for the homography calculation. To compute the homography, I used least squares, calculating Ah = b, where A is a n x 8 matrix, h is a 8 x 1 matrix of parameters we wish to solve for, and finally b is an nx1 matrix of points, with the x coordinates stacked on top of the y coordinates.

Part 3: Image Rectification

Original Image of Laptop
Rectified laptop
Original Image of Basket
Rectified basket

Part 4: Image Warping

Attempted to stitch and warp the images using inverse warp.

Image Warping Attempt
Image Warping Attempt

Part 5: What I learned

The coolest thing I learned in this project is image rectification. It was really cool to create "better looking" images and really understand the mechanism of inverse warping and interpolation.