ICSA based projection pursuit clustering with LDA index and its application in SAR image segmentation

The performance of the traditional clustering algorithm is not always satisfied with the high-dimensional datasets, which make clustering method limited in many application. To solve this problem, Projection Pursuit with LDA index based on Immune Clonal Selection Algorithm is proposed in this paper. Projection pursuit strategy can maintain consistent Euclidean distances between points in the low-dimensional embeddings where the ICSA is used to search optimal projection direction. In the proposed algorithm, k-mean clustering algorithm is used to partition the reduced data and give labels to LDA index, and iterations have been done for several times can converge quickly to obtain a clustering result. The experimental results on UCI data show that the present method has better clustering accuracy compared with traditional linear dimension reduction method for principle component analysis (PCA). And the experiments on SAR image segmentation also have a good performance.