Fundamental Matrix Estimation Based on Improved Genetic Algorithm

摘要

Fundamental matrix estimation is the key technology in 3D reconstruction and widely used in many aspects in computer vision. In this paper, a global search genetic algorithm combining with a local search hill climbing algorithm is proposed to optimize MAPSAC algorithm for estimating fundamental matrix. The average distances between points and epipolar lines under noise and outliers are investigated with synthetic and real image feature point data respectively. The simulation results show proposed method is more robust to noise and outliers and can estimate fundamental matrix more precisely.