A Variance Threshold and Clustering Based Approach for Automatic Marine Target Recognition for Synthetic Aperture Radars

This paper proposes a method based on a variance threshold based clustering method for automatic target recog-nition (ATR) of marine targets for synthetic-aperture-radar (SAR) technology. The proposed method forms clusters of radar reflectivity data, subsequently followed by bounding the clusters. Finally, from the bounded clusters, the targets are detected based on an estimation algorithm with input, the variance of each cluster. For clustering, we applied standard K-means clustering method. The performance of the proposed method is further studied for its performance for different noise-levels.