摘要
本文介绍了基于ＢＰ神经网络的遥感模式识别方法的特点，同时引入分形维数的概念及其在图像中的计算和应用，并将分维作为反映各类别数据纹理特征的附加波段分量，加到网络的输入层，明显地提高了分类精度。
This paper has presented a classification method of remotely-sensed Ima-ge data based on BP neural network,and introduced the concept of fractal dimensionwhich describes image texture suitably;The approach is also given for calculatingfractal dimension applied in image field.Finally,fractal dimension is adopted astbe additional sector of input vectors in neural network,which has greatly improvedclassification accuracy.
出处
《环境遥感》
CSCD
1994年第1期68-72,共5页
关键词
遥感
分形维数
神经网络
模式识别
Remote sensing,Fractal dimension,BP neural network
