XE33PVR::Lab3 - 2D Homography - advanced

Introduction

The goal of the lab is to learn how to map pixels from one image onto another image using a 2D homography from point correspondences.

Image transformation

Download the following images to your local directory:

Estimate image homography H from 4 correspondences selected manually. Use functions maketform and imtransform to transform the first image.
Replace the relevan part of the second image by transformed first image. Display differences in the original and new images (for example put intensities of one image in red channel and intensities of the other image into a blue channel).
Instead of using only 4 manually selected points, use the web WBS demo to obtain multiple correspondences and repeat the process. What is the difference in results?
Think of ways to improve the visual quality of the results. Get inspired by harris.m , nonmaxsuppts.m, and Normalized Cross Correlation.


[XE33PVR labs | Responsible: Ondrej Chum ]
Last modified 19 Oct 2008.