Summary

In this chapter, the aim was to view computer vision from a geometrical point of view. Starting with understanding how an image is formed using a pinhole camera, there was a discussion on how to incorporate three-dimensional worlds using multi-image formation. We saw an explanation of Visual Odometry with an introduction to Visual SLAM. The various steps involved in SLAM were explained and a demo of using ORB-SLAM was also shown, so that we could see a SLAM operation as it happened. This is basic motivation to extend the SLAM solution for various other datasets, and so create interesting applications.