基于遗传规划的多光谱图像分类

针对多光谱图像分类这一多类别模式识别问题,将二进制纠错编码与GP(GeneticProgramming)算法相结合,并用改进后的编码矩阵代替原先的二进制编码矩阵对图像进行分类,从而建立了新的基于GP的多光谱图像分类算法,给出了用该方法对多光谱图像中地物进行分类的实例。结果表明与以往基于GP的分类方法相比,该文方法体现出较高的分类性能,为遗传规划在多类别模式识别问题中的应用提供了又一条可行的途径。

Classification of Multispectral Data Using Genetic Programming

Genetic Programming(GP) can discover relationships among abserved data and express them mathematically.However,some existed algorithms based on GP can not resolve the multi-class classification issues accurately.This paper develops a new GP-based method for multi-class classification problems,in which,authors combine the binary error-correction codes and GP,and use an improved coding matrix.Remote multispectral images are used to examine the method by comparing the performances of several algorithms in table and figures.The results show that the approaches are able to produce classifiers more accurately than traditional algorithms.