3D Reconstruction using RGB images
In this project, our task was to reconstruct an object in 3D using 3D rgb images.
In this project we implemented two different methods i.e. the 8-point algorithm and the 7-point algorithm to estimate the fundamental matrix from corresponding points in two images. Given the fun- damental matrix and calibrated intrinsics we computed the essential matrix and used it to compute a 3D metric reconstruction from 2D correspondences using triangulation.
The performance was evaluated on the Middlebury multi-view dataset.
The view from two different camera poses.
Scatter plot of 3D points extracted from the above images.