Fast Method for Two-dimensional Renyi’s Entropy-based Thresholding

Abstract—Two-dimensional (2-D) thresholding can give a better segmentation than one-dimensional thresholding by taking the spatial correlation of the image. Unfortunately, the computational cost is an obstacle for the implementation of real-time image processing. In this paper, a fast method for 2-D Renyi’s entropy-based thresholding, which is the generalized method of Shannon entropic method and the correlation entropic method, is proposed. In order to diminish the processing time required for calculating 2-D histogram, a fast scheme is introduced and the computational complexity is reduced from O(W 2MN) to O(MN). More importantly, based on the proposed method, the computational complexity for selecting the optimal threshold value is reduced from O(L4) to O(L2). The effectiveness of this method is illustrated by experimental results