Extenic Image Classifier and Its Application in the Land Use Classification

Abstract

The theory of Extenic was put forward by the Chinese scientist, Prof. Cai wen, in 1983. As a new kind of artificial intelligence method, it has been used in the many areas, this paper use it to extract the land use information from the remote sensing image. The research area in this paper is the plan area of city of the Du Jiang Yan, which locates in the northwest of the Sichuan basin and the data used here is the HJ satellite image. It has four kinds of land use type in the ground, the forest land, the water body, the built-up land and the farm land, different land use type in the ground has different color in the false color image, and the 11 subtypes are divided according to the relationship analysis of the land use type and the pixel color of the image. The 11 standard matter-element models, which corresponding to 11 subtypes of the land use, are built at first, then calculate the extension correlation degrees of each pixel in the image to the 11 standard matter-element models and finally, and finally the pixel is determined to belong the land use type which has the biggest extension correlation degree. The right rate of the classification is about 87.2% and the Kappa index is 0.86, it shows that the classifier based on the theory of the Extenic has the high precision in the classification of the remote sensing images.