Image Classification Of Remote Sensing And Its Application Based On LVQ Neural Network And Gray Level Co-occurrence Matrix

In remote sensing image classification process, the general use of remote sensing spectral imaging characteristics of information as classified, but the complexity of the surface in some regions, remote sensing spectral characteristics of the image spectrum and the synonym is a serious problem with the spectrum of foreign bodies, various types of spectral features can be divided into and well, the existing classification accuracy is also not satisfactory. With its own development of remote sensing image change, the requirements of the classification accuracy has been improving, to that end, scholars from both home and abroad to improve the classification accuracy of ways: First, improve the existing algorithms, to find new algorithms, such as artificial classification neural network, rough set classification, the expert classification, etc.; Second, auxiliary features and a combination of spectral features.This article uses remote sensing imagery processing software ENVI4.3 the wool you to cover the area ETM image to the Minjiang River upstream to carry on the cluster, then uses the LVQ network in this foundation to carry on separately to it classifies once more, takes the different implicit function node to carry on to the network tests and chooses the most superior LVQ neural network model to assign the region to carry on the classification, finally carries on the appraisal to the classified result.In order to further increase the classified precision, this article has a texture characteristic based on the 9*9 gradation paragenesis matrix, mainly carries on the classification using the LVQ neural network to the remote sensing image, simultaneously also uses BP, the RBF network conducts the contrast research to the remote sensing image classification and with the LVQ network classified result, the experiment indicated auxiliary by the texture characteristic classified method which produces by the gradation paragenesis matrix is effective, is helpful in the enhancement classification precision.