PROJECT TITLE :

Unsupervised Band Selection Based on Evolutionary Multiobjective Optimization for Hyperspectral Images

ABSTRACT:

Band choice is a vital preprocessing step for hyperspectral image processing. Many valid criteria have been proposed for band choice, and these criteria model band selection as one-objective optimization problem. During this paper, a unique multiobjective model is first built for band choice. In this model, 2 objective functions with a conflicting relationship are designed. One objective function is ready as information entropy to represent the information contained in the selected band subsets, and the other one is set as the amount of selected bands. Then, primarily based on this model, a replacement unsupervised band choice technique referred to as multiobjective optimization band selection (MOBS) is proposed. In the MOBS methodology, these two objective functions are optimized simultaneously by a multiobjective evolutionary algorithm to search out the most effective tradeoff solutions. The proposed method shows 2 unique characters. It can acquire a series of band subsets with totally different numbers of bands during a single run to supply additional choices for call manufacturers. Moreover, these band subsets with totally different numbers of bands can communicate with each alternative and have a coevolutionary relationship, that means that they can be optimized in an exceedingly cooperative way. Since it's unsupervised, the proposed algorithm is compared with some related and up to date unsupervised ways for hyperspectral image band selection to evaluate the quality of the obtained band subsets. Experimental results show that the proposed method can generate a group of band subsets with different numbers of bands in a single run which these band subsets have a stable smart performance on classification for different data sets.