Threshold Selection on Circular Histogram Using Renyi Entropy

Abstract

Using the circular histogram of the H component in the HSI color model for threshold selection is a way of color image segmentation. The maximum Shannon entropy thresholding on the circular histogram is an effective segmentation method. Considering that Renyi entropy is one of the generalized forms of Shannon entropy, this paper extends the maximum Shannon entropy threshold selection method on circular histogram to Renyi entropy case, gives recursive algorithm to reduce the time complexity of Renyi entropy threshold selection, and discusses the determination of parameters in Renyi entropy threshold selection. The experimental comparison effect shows that the maximum Renyi entropy threshold selection method outperforms the maximum Shannon entropy threshold selection method.