In this project we inplement Matlab code to estimate camera calibration, specifically estimation of camera projection matrix, and fundamental matrix. We have Performed accurate estimation of camera projection matrix and the fundamental matrix can each be estimated using point correspondences related by epipolar lines on both the images. We have used linear regression to estimate the matrices. We have used RANSAC in conjuction with fundamental matrix to deal with outliers.

1. The code uses Matlab library vlfeat. vlfeat not included in the submission. The user MUST download vlfeat from http://vlfeat.org in order to successfully run this project.

2. For better performance, normalization has been performed on the matching interest points for Fundamental Matrix estimation. A function named Normalized_estimate_fundamental_matrix implements this.

3. The function estimate_fundamental_matrix implements estimation of fundamental matrix without normalization of matching points.

4. To implement Normalized_estimate_fundamental_matrix, uncomment the function call in proj3_part2.m and comment out the function call to estimate_fundamental_matrix.

5. To implement estimate_fundamental_matrix, uncomment the function call in proj3_part2.m and comment out Normalized_estimate_fundamental_matrix function all.

6. proj3_part3.m calls ransac_fundamental_matrix which in turn uses Normalized_estimate_fundamental_matrix. There is no need for the user to modify the function call here.

7. The threshold remains same for all the images.

The results and visualizations are presented in a html page found here
http://htmlpreview.github.io/?https://github.com/anishagartia/fundamental-matrix-estimation/blob/master/html/index.html