论文部分内容阅读
本文将遗传算法(GA)应用于非监督训练,提高了遥感数据的分类精度。遗传竞争学习算法(GA-CL)综合了遗传算法和简单的竞争学习算法,可用于改进非监督训练的结果。遗传算法在典型样本聚类的过程中可以避免得到局部最优值。Jeffries-Matusita(J-M)距离法是通过统计测量两个训练类别之间的分离度,可用于评价这种算法。将此算法应用于TM数据的结果显示,遗传算法改进了简单的竞争学习算法,与其他非监督训练算法相比,其提供了K-均值,GA-K-均值和简单的竞争学习算法。
In this paper, Genetic Algorithm (GA) is applied to unsupervised training to improve the classification accuracy of remote sensing data. Genetic Competitive Learning Algorithm (GA-CL) combines genetic algorithms and a simple competitive learning algorithm that can be used to improve the results of unsupervised training. Genetic algorithm can avoid getting local optimal value in the process of typical sample clustering. Jeffries-Matusita (J-M) The distance method is a measure of the degree of separation between two training categories by statistics and can be used to evaluate this algorithm. The results of applying this algorithm to TM data show that GA improves the simple competitive learning algorithm, which provides K-means, GA-K-means and simple competitive learning algorithm compared with other unsupervised training algorithms.