一种改进的遗传算法在图像匹配技术中的应用  

Application of an Genetic Algorithm in Image Matching Technique

作　　者：王曼[1] 王魁生[1] 

机构地区：[1]西安石油大学计算机学院,陕西西安710065

出　　处：《计算机仿真》2008年第3期86-89,共4页Computer Simulation

摘　　要：为了提高导航精度,研究应用于导航系统中的图像匹配技术。针对传统的遗传算法全局寻优能力不高这一局限性,提出了一种改进的遗传算法。算法在选择操作中改进了最优保存策略。通过引入群体的平均适应度,将选择的重点定位在适应性好的群体而不是个体上,减小了算法陷于局部极值的可能性,提高了搜索效率。选取1.6GHz的工艺机作为实验环境对算法测试50次,测试结果平均匹配精度为92.7%,且127ms的平均搜索时间远少于传统算法,验证了算法的可行性。The application of image matching technique in navigation system is studied to increase the navigation accuracy . An advanced genetic algorithm is proposed because the conventional genetic algorithm has lower global optimization capability. This algorithm improves Elitist Model in selection - operation. By putting the point of choice on community of good adaptability rather than on an individual, it reduces the possibility that the algorithm could only get the local extremum. Also, it could increase the searching - efficiency. Testing this algorithm by 50 times, it could be found that the average matching - accuracy is 92.7% and the average searching time of 127ms is lower than traditional algorithm. So the algorithm is feasible.

关 键 词：遗传算法 图像匹配 模板 全局寻优 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]