Research On The Classification Of Remote Sensing Images Based On ACO

In general, the theme of this thesis that remote sensing image classification based on data mining by ant colony optimization is discussed from three aspects.First, introducing the general methods and characteristics of classification issue. Second, introducing the basic principles, mathematical models and processing procedure of ant colony optimization. Finally, the problem have to solve that aco is applied to the remote sensing image classification.About remote sensing image classification, using the latest data of Landsat-8,discuss the problems about texture transformation, vegetation transformation, principal component analysis, independent component analysis, terrain factor extraction and multiband optimum index factor extraction. The spectrum, remote sensing index, texture, terrain and the characteristics of the linear transformation combined into a multichannel file, as initial features set of classification, its feature selection scheme is given.On ACO, taking the double bridges experiment as the quotes, discusses the principle, the mathematical model and algorithmthe processing for traveling salesman problem and datamining based on ACO,mainly introduces the Ant-Miner model. Introduce a new improvement scheme on the basis of summarizing the general framework of ACO.Conjunctively, discuss the problem about data discretization, rule construction, rule pruning, pheromone and inspiring. Finally, an example is designed to compare the ACO method and ML classification results, and draw the conclusion.