A Tsallis-Entropy Image Thresholding Method Based on Two-Dimensional Histogram Obique Segmentation

Image Segmentation is one of important tasks in conventional or document image processing. Tsallis-entropy based image thresholding method has been considered one of the most efficient ways for image segmentation. At present one useful way in entropy-thresholding segmentation is based 2D vertical segmentation, but obvious limitations exist in this approach. In this paper, a method 2D obique segmentation is applied. Meanwhile motivate by Kullback-Leibler (KL) distance which measures the information discrepance between two different sources, a new optimization based on minimizing imitational Kullback-Leibler distance (MIKL) criterion function is proposed. It is a new technique that image threshold value by 2D Tsallis-entropy obique segmentation based on MIKL. Some typical results show the superiorty of the proposed technique over references, achieving better effect wherever in the document or conventional image segmentation.