Multilevel Quantum Evolutionary Butterfly Optimization Algorithm for Automatic Clustering of Hyperspectral Images.

Hyperspectral images contain a huge amount of spectral and spatial information. Automatic clustering of such images is a difficult task. Two algorithms viz., Qubit Evolutionary Butterfly Optimization algorithm and Qutrit Evolutionary Butterfly Optimization algorithm, are proposed for the automatic clustering of hyperspectral images. Implementing quantum mutation operators enhances the exploration phase and prevents it from getting stuck in local optima, a drawback of their classical version. This is the first time quantum versions of the Butterfly Optimization Algorithm are proposed and implemented on such a large dataset. The two proposed algorithms are compared to the classical Butterfly Optimization Algorithm and classical Evolutionary Butterfly Optimization Algorithms, and applied to the Xuzhou HYSPEX dataset. Statistical tests like mean, standard deviation, One-Way ANOVA test, and Tukey’s Post Hoc test are performed on all the algorithms to examine their efficiency. The Correlation Based Cluster Validity Index is used as the fitness function. The quality of clustering is determined using the F, F’, and Q scores. The qutrit version is found to produce more optimal results, followed by the qubit version. The qutrit version converges faster with better optimal results by using a smaller population compared to the other three algorithms.