I am a postdoctoral researcher at the Mathematical Imaging Group (MIG), Lund University since October 2010. The overall aim of computer vision research is to make computers "see" and this can refer to any high-level vision-based activity such as recognizing faces, people, objects or using vision to navigate and avoid obstacles.

My current research is on 3D reconstruction and mapping from video sequences, the so called structure from motion problem. In particular I work with developing fast and numerically stable methods for solving systems of polynomial equations arising from such problems. More recently I also work on large scale bundle adjustment using iterative methods such as conjugate gradients.

Below are two pictures from things I've played with: A point-cloud reconstruction of a toy dinosaur using a triangulation technique we are currently working on (left) and a sudoku puzzle which was solved by a vision based sudoku solver written in Matlab (right).