Study Of Computer Classification Of Remote Sensing Images

With the rapid development of remote sensing technology and information techniques,the Automatic extraction of image informations from remote sensing images has become a means of interpretation of remote sensing images.But the accuracy of computer classification is lower, it can not deal with the map of larger-scale and medium-scale.Therefore, the research of automatic extraction from image is an important direction for future development. Automatic recognition of remote sensing image is a great challenge in the field of remote sensing, computer vision,vague recognition and so on.Remote sensing image classification is the technical process in which the image's pixels are divided into a number of categories.In this paper, the study of remote sensing image classification has adopted two ways: one is based on the spectral characteristics and the other is based on the non-spectral characteristics, an auxiliary classification method. For the first one, it is a way that merge the traditional unsupervised classification methods and supervised classification methods.A major study of the supervised classification algorithms and unsupervised classification algorithms, the algorithms for the classification of each tested also made a detailed analysis of the corresponding classification.For the non-spectra characteristics supporting the classification of remote sensing images,in this paper, it is based on the gray level co-occurrence matrix to extract the texture images.Then put the image texture classification in the form of spectral characteristics classification. The classification results with the texture images are analyzed and compared with the traditional ways.In the process of extracting texture images vc program has taken.The results showed that the texture image's participation to some extent increase the classification accuracy of remote sensing images.