Hyperspectal image clustering using ant colony optimization(ACO) improved by K-means algorithm

Based on the comparison of K-means algorithm and ant colony optimization (ACO) algorithm in image clustering, this essay proposed a K-means-ACO algorithm to solve the problem of misclassification of K-means and slow convergence of ACO. K-means-ACO algorithm takes the results of K-means as the elicitation information of ACO, which adds illumination probability and illumination pixels in ants seeking rules of ACO, permits ants select nodes according to pheromone concentrations directly instead of probability, makes the elicitation information can be fully without altering the random search quality of ACO. Through the verification of simulation data and real data, the K-means-ACO algorithm can improve the clustering accuracy for adjusting the misclassification of K-means, and improve the ACO´s convergence speed.