A Clonal Selection Algorithm for Classification of Mangroves Remote Sensing Image

Abstract

As a new computational intelligence model based on artificial immune systems, clonal selection algorithm has been widely utilized for data analysis and pattern recognition. Recently it was applied to remote-sensing image classification. However, due to the similar spectral feature between mangroves and other land cover types such as agricultural land and forests, serious misclassification and confusion can develop in mangroves classification using conventional methods. This paper proposes a clonal selection based supervised classification algorithm which takes into account not only spectral feature but also geographical feature and image feature. The proposed algorithm searches the best cluster centers for various types of training samples by the improved clonal selection algorithm. The antibody represents the candidate solution, while antigen is reflected by affinity function. The antibody is encoding by decimal way. The inner superiority and the outer superiority together are used to measure the superiority of antibody. The selection operator and mutation operator are designed to guarantee the diversity and global optimality. Experiments are performed and the results show that the proposed method can improve the extraction accuracy of mangroves effectively.