Evaluating the Potential of Particle Swarm Optimization for Hyperspectral Image Clustering in Minimum Noise Fraction Fea

Particle Swarm Optimization methods are the optimization techniques inspired by the social movements of animals. In these methods, the particles movements are based on simple rules, but make complex overall behavior and search of the space. The clustering

  • PDF / 256,432 Bytes
  • 11 Pages / 439.37 x 666.142 pts Page_size
  • 24 Downloads / 190 Views

DOWNLOAD

REPORT


Evaluating the Potential of Particle Swarm Optimization for Hyperspectral Image Clustering in Minimum Noise Fraction Feature Space Shahin Rahmatollahi Namin, Amin Alizadeh Naeini, and Farhad Samadzadegan

Abstract Particle Swarm Optimization methods are the optimization techniques inspired by the social movements of animals. In these methods, the particles movements are based on simple rules, but make complex overall behavior and search of the space. The clustering is the process of dividing the existing data to diverse groups based on the inherent characteristics and similarity of them and can also be seen as an optimization problem. Due to the complexity of this problem in hyperspectral remotely sensed data, different feature space and clustering techniques are applied and evaluated in this field. In this paper, a PSO clustering algorithm is evaluated in Minimum noise fraction space for hyperspectral AVIRIS image taken over the northwest indiana’s indian pine agricultural land. The comparison of the results with the K-means clustering method shows better obtained performance for the PSO clustering in minimum noise fraction feature space.

7.1

Introduction

The Particle Swarm Optimizations (PSO) are population-based tools with high potentiality to solve complex optimization applications [1]. These algorithms have global and local search mechanisms to reach to the optimum solutions iteratively. Currently there exists a great amount of literature investigating different aspects of these algorithms and evaluating their potential to solve complex optimization applications. One of the complex applications that the optimization algorithms can be applied is clustering. Clustering is a method of partitioning a set of data into subsets, in a way that the data grouped together, are similar to each other and are diverse from the data in

S. Rahmatollahi Namin (*) • A.A. Naeini • F. Samadzadegan Department of Surveying Engineering, College of Engineering, University of Tehran, Tehran, Iran e-mail: [email protected]; [email protected]; [email protected] A. Madureira et al., Computational Intelligence and Decision Making: Trends and Applications, Intelligent Systems, Control and Automation: Science and Engineering 61, DOI 10.1007/978-94-007-4722-7_7, # Springer Science+Business Media Dordrecht 2013

69

70

S. Rahmatollahi Namin et al.

other clusters [2]. In this study, a PSO optimization algorithm is used in order to find the best cluster centers in a way that the accuracy of the clusters are optimized. Despite the inherent complexities of the clustering that the algorithms should be dealing with, the clustering of the hyperspectral images bring other necessities which are the result of the high dimensionality and correlation of these data. Although the high spectral resolution in these images allows recognition of spectrally similar objects, the high dimensionality of these data and the resulting correlation between them decreases the performance and accuracy of the common clustering algorithms, because a great number of loc