Plane Identification and Segmentation from Unclaibrated Stereo Based on Evolutionary Approach

Abstract

Planes are widely used in various vision tasks due to their abundance especially in man-made environments. Assuming an uncalibrated stereo image acquisition system, we propose a new plane identification and segmentation algorithm using evolutionary approach. The proposed algorithm enables faster and more robust estimation of the plane since it progressively search the best solution in a guided direstion. Experimental results show that the proposed algorithm has higher performance than the existing method. © 2004 IEEE.