The method research of city vegetation information abstraction based on high-resolution IKONOS image

In the first part of this article, traditional methods for deriving city vegetation information through remotely sensed images, the study of the area's characteristics and the research flow are introduced. In this article, two main issues are studied, one is city shadow correction, the other is the classifying method of city vegetation through remote sensing images. The biggest problem of high-resolution images is the shadow of high buildings. City shadows decrease the classification accuracy, so a shadow adjustment method must be studied. In this article, we analyze the radiation correction model, and acquire the shadow adjustment model. The model's parameters can be calculated through the image pixel values. The result of shadow correction shows that the model can correct for city shadow. After shadow correction, the vegetation's pixel value in the shadow is similar to the pixel value of vegetation in other areas. Shadow correction increases classification accuracy. In the second part of this article, three methods are studied to derive city vegetation information, including the vegetation index method, a back propagation neural network method, and a texture method. Finally, the three methods' classification accuracies are calculated and appraised. A conclusion is drawn, which is that the texture classification method is a good classification method. The accuracy of texture classification method can reach 91.53%.