A fast two-class classifier for 2D data using complex-moment-preserving principle

A new moment-preserving classifier for two-class clustering is suggested. Based on preserving the complex moments of two-dimensional (2D) input data, an analytic, non-iterative and unsupervised classifier is proposed. This new classifier is suitable for applications requiring fast automatic two-class clustering of 2D data or fast automatic hierarchical clustering. Furthermore, the computation time is of order of data size and hence much faster than the well known iterative k-means algorithm. Experimental results show that the proposed classifier can acquire acceptable clustering results.