Abstract:

In order to improve water quality evaluation of multi-spectral image accurately, this paper puts forward a model for water quality evaluation of Radial Basis Function(RBF) neural network based on optimized parameters of Genetic Algorithm(GA). The method uses SPOT-5 data and the water quality field data. It chooses four representative water quality parameters. RBF neural network is trained and tested, and the parameters of RBF neural network are optimized by particle swarm optimization algorithms. Water quality parameters of COD, NH3-N, DO, CODmn are retrieved by the trained RBF neural network. Experimental result shows that the method has more accuracy than the routine method.

Key words: Radial Basis Function(RBF) neural network, Genetic Algorithm(GA), optimized parameter, high-resolution remote sensing image, water quality evaluation

摘要：

为进一步提高多光谱图像水质反演的评价精度，提出一种基于遗传算法(GA)优选参数的径向基函数(RBF)神经网络水质评价方法。利用高分辨率多光谱遥感SPOT-5数据和水质实地监测数据，得到符合条件且具有代表性的4类水质变量，对RBF神经网络进行训练和测试，用遗传算法对RBF神经网络的参数进行优化。在训练好的RBF神经网络模型基础之上对COD、NH3-N、DO、CODmn水质参数进行反演。实验结果表明，该水质反演模型较常规的方法有更高的反演精度。

关键词: 径向基函数神经网络, 遗传算法, 优选参数, 高分辨遥感影像, 水质评价

CLC Number: 

• TP391