Research And Application Of Remote Sensing Image Classification Based On Partial Binary Tree Twin Support Vector Machine

Remote sensing technique is an emerging integrated technology which has developed since the 1960s, it is closely related to other numerous sciences and technologies. And it is also one of the most effective technologies and methods to study the. resources of the earth. Identifying the category of the ground object by analyzing remote sensing images is a very important topic in the study of remote sensing.The main task of remote sensing image classification is to identify and estimate the category and distribution of the objects on the earth on the basis of characteristics of the electromagnetic radiation information of ground objects acquired by the sensor working far away from the ground platform.According to the requirements of the project "state grid disaster warning platform of the transmission lines", which I have took part in during my graduate study, we study on the remote sensing image classification and develop a complete system of the remote sensing image classification in the thesis. The specific content of the work and research results are as follows:(a) We study some classical methods of remote sensing image classification, including remote sensing image classification based on Maximum Likelihood, remote sensing image classification based on Iterative Self-organizing Data Analysis Techniques Algorithm. This study helped me to understand the process of remote sensing image classification, and get basic knowledge of the difficult in traditional remote sensing image classification, which is the foundation for the further research of remote sensing image classification method.(b) We research Support Vector Machine method in the application of remote sensing image classification, which focuses on the Twin Support Vector Machine and Binary Tree Support Vector Machine. According to the advantages of these two methods, we proposed a combined classification method of remote sensing images based on Partial Binary Tree Twin Support Vector Machine.(c) We develop a set of remote sensing image classification system based on the original intention of this thesis. The classification results of this system meet the demand model of remote sensing data classification prediction of fire model in the project "state grid disaster warning platform of the transmission lines".