自组织神经网络的土地覆盖图像识别分析应用

土地利用/土地覆盖数据的识别,采用传统的方法进行类型识别运算量大,精度也不太理想。本文重点讨论采用自组织神经网络方法,并在MATLAB平台下对其算法进行了实现,最后将分类识别结果与最大似然法分类结果进行了精度比较分析。结果表明,其分类精度明显高于最大似然法的分类精度,是一种土地覆盖图像识别分类的有效方法。

Studies on Application of Self-organizing Artificial Neural Network in Remote Sensing Classification Recognition of Land Cover

Remote sensing technology has already been an important means of LUCC(Land Use & Land Cover). When the traditional methods are used in the land cover classification recognition, the large amount of operation is needed and the accuracy is not perfect. This paper mainly discusses the method of self-organizing neural network that has been very effectively applied to the remote sensing classification processing of land cover and presents its classification algorithm implemented by MATLAB developing flat. Finally, its accuracy of classification recognition is compared to the Maximum Likelihood Classifier(MLC). The result shows that the classification accuracy of self-organizing network method is obviously higher than that of the traditional MLC method, so it is an effective land cover classification method in remote sensing field.