摘要
提出遗传算法(简称GA)与单纯形法组合的影像纹理分类方法(简称GASPX)。单纯形法是一种局部搜索方法,它通过反射,扩张,收缩操作,求得新的单纯形点,组成新单纯形,新单纯形比前一个单纯形更接近局部最优解。这种寻优方法收敛速度快,它与GA组合起来可以改善单独使用GA收敛速度慢的缺陷。由于在组合算法中是多个单纯形的局部区域的并行搜索,避免GA优化过程中过早收敛于局部最优解的现象出现。通过5种不同类别航空影像纹理识别的试验,并与GA的结果作对比,结果表明GASPX法优于GA法。
Propose a new method to classify image texture by combination of Genetic algorithm and Simplex method(GASPX).The Simplex method is a local searching method that gets the new simplex points by reflection,expansion and contraction.Then we can get new simplex through the new simplex points.The new simplex is closer to the local optima than the previous simplex.The method converges quickly and can amend the defect that converges slowly of GA when combined with GA.We can avoid early convergence in the course of GA optimization through using many simplex searches of the optima of local area in parallel in the combinatorial algorithm.Experimental results show that GASPX is better than GA through the trial of recognition 5 kinds of aerial image texture and comparison with the result from GA.
出处
《测绘学报》
EI
CSCD
北大核心
2003年第4期325-329,共5页
Acta Geodaetica et Cartographica Sinica
基金
国家自然科学基金资助项目(40271094)
关键词
遗传算法
单纯形法
影像纹理分类
优化
genetic algorithm
simplex
classificatiion of image texture
