Study of Fundamental Matrix High-Precision Estimation Based on Particle Swarm Optimization

The paper researches the issue that fundamental matrix is recovered in images. Firstly, the corners are detected by Harris operator, and matches cross-correlation corners. According to the definition of polar equation, the polar of feature point must pass corresponding point in matching images. Then, to extract correctly matching points and eliminate error by the criterion which is the distance between corner and corresponding polar. At the end, More precise fundamental matrix is got by Particle Swarm Optimization according to the target function that is residual error of polar equation. Experimental result proves that interrelated arithmetic is effective.