An Algorithm for Estimating Fundamental Matrix Based on Removing the Exceptional Points

In this paper, an algorithm for estimating fundamental matrix based on removing the exceptional points is proposed. It uses normalized 8-point algorithm to estimate the initial value of the fundamental matrix, and then refines it by eliminating the exceptional points. Furthermore, the genetic algorithm (GA) is used to optimize the initial fundamental matrix by minimizing the Sampson error. Experiment results show that our algorithm is quite correct and reasonable for estimating the fundamental matrix.