I developed a 3D reconstruction algorithm as part of a computer vision course. The goal was to reconstruct a scene in 3D using two successive images from the KITTI Vision Benchmark. The project was developed on Matlab with the Image Processing Toolbox and Peter Kovesi's toolbox (http://www.peterkovesi.com/matlabfns/index.html).
Algorithm :
- Interest points detection using the Harris detector on both images
- Matching by correlation of the detected points
- Robust estimation of the fundamental matrix
- Projection matrices reconstruction in the two images
- Computation of the 3D structure by triangulation