Feature Selection and Classification Based on Ant Colony Algorithm for Hyperspectral Remote Sensing Images

Abstract

This paper proposes a method of feature selection and classifcaition based on ant colony algorithm for hyperspectral remote sensing image. After all features are randomly projected on a plane, each ant stochastically selects a feature on the plane firstly, and then decides which route to be selected in terms of the criterion function among features. Whereafter the feature combination is formed. At last, using combination feature, the classification of AVIRIS image is carried out by maximum likelihood classifier. In order to verify the effectiveness of this algorithm, the approach is compared with the classical suboptimal search technique, using AVIRIS images as a data set. Experimental results prove the processing that based on ant colony algorithm is more effective and is fit for the band selection of hyperspectral image.

Identify important spectrum bands for classification using importances of wrapper selection applied to hyperspectral data

“…Variants such as Sequential Forward Floating Search (SFFS) or Sequential Backward Floating Search (SBFS) [21], or Steepest Ascent (SA) [6] have also been proposed. Several stochastic optimization strategies have also been used for feature selection, including Genetic Algorithms (GA) [12,10,19,13], Particle Swarm Optimization (PSO) [7], clonal selection [14], ant colony [22] or even simulated annealing [5]. Thus, many FS methods have been proposed in literature.…”

Wrapper based feature selection in hyperspectral image data using self-adaptive differential evolution

“…Different suboptimal feature selection techniques have been applied in hyperspectral images, such as SFS, SBS, SFFS, and SBFS [2], [15]. Most of the traditional techniques have limitations in suboptimal subset selection for hyperspectral images due to strong correlation between bands [9], [12]. Other technique for subset generation is based on stochastic optimization techniques like evolutionary based or swarm based search techniques.…”

“…In hyperspectral images, supervised feature selection methods [5], [6], [7], [8], [9], [10], [11], [12], [13] have two main steps: subset generation and subset evaluation. Subset generation is essentially a process of heuristic search, where each point in the search space specify a candidate subset for evaluation.…”

“…In HPSOGA-SS, all processes are the same as HPSOGA except a fine tuning of particles after each generation by sequential search method. ACO based methods use ant colony algorithm for searching technique and use wrapper based method [9] or filter based method [12] for calculating evaluation measure.…”

“…There are various stochastic optimization based feature selection techniques for hyperspectral images like genetic algorithm (GA) [8], [10], [11], [13], particle swarm optimization (PSO) [5], [6], [7], ant colony optimization (ACO) [9], [12] etc. Genetic algorithm based feature selection methods for hyperspectral images use binary GA and evaluation measure is calculated by either filter method [8] or wrapper method [10], [11].…”