遗传算法在立体图象匹配中的应用

基于传统立体图象匹配技术，提出了一种新的使用遗传算法的特征匹配方法．实验中，采用了两种特征提取算法，以适应不同图象要求．实验用图包括自然景物图和人工合成图．此外，用ＢｒａｎｃｈＡｎｄＢｏｕｎｄ算法得出比较结果．实验结果表明，遗传算法有好的匹配正确率和收敛速度，具有鲁棒性，适合立体图象匹配．

Application of Genetic Algorithms in Stereo lmage Matching

This paper introduces a new approach of feature matching using genetic algorithms. Based on the traditional processing technology in stereo image matching, this algorithm is devised to be sui table for stereo image matching. Two ways of feature selecting are used in the experiment to adapt to different images, which include natural and synthetic ones. We get the comparative results using branch and bound algorithm. From the results of experiment, it is shown that the genetic algorithm has high correctness and speed. The genetic algorithms can be used in stereo image matching and have the robust property as the theory shows.