摘要
本文介绍了几种基于人工神经网络的土地覆盖/土地利用分类方法,对其优缺点进行了评述,并提出了一种基于遗传BP算法的土地覆盖/土地利用分类方法。这一算法克服了BP算法收敛速度慢、易陷入局部极小等缺陷,能科学地选取训练样本,提高分类精度。
This paper introduces several classification methods for land cover/land use based on artificial neural network, and gives some comments on their advantages and disadvantages. The authors propose a new classification method for land cover/land use based on the integration of genetic algorithm and back - propagation algorithm. It overcomes the limitations of the back ?propagation algorithm in slow convergent rate and getting into local minimus. In addition, it can select training samples scientifically.
出处
《湖北地矿》
2002年第2期31-37,共7页
Hubei Geology & Mineral Resources
