Sparse resultant-based methods with their applications to generalized cameras

Finding the solutions of a set of equations is a classical problem in mathematics, with a rich literature. Recently, these mathematical techniques have gained popularity in various fields of computer science, specifically computer vision and robotics. This thesis proposes novel algorithms to efficiently solve the equations, specifically arising in computer vision. The algorithms borrow concepts from algebra and geometry, and are implemented as easy-to-use software programs. Further, the thesis also studies the challenging problems involving multi-camera systems and solves them using the proposed algorithms. The multi-camera systems are important in applications such as autonomous driving, SLAM (simultaneous localization and mapping), 3D reconstruction, etc.