摘要
针对Otsu阈值法仅适合图像目标和背景分布呈正态分布且两者像素量和方差要求基本相当的不足,提出了基于Itakura-SaitoS散度的图像阈值化分割新算法。实验结果表明,对有些图像采用Itakura-Saito散度的阈值法比传统Otsu阈值法能够获得较好的分割效果。
Aimed at the shortage of Otsu' s thresholding method which is only suitable for the probability models of image target and background having normal distribution, and the pixel amount and variance of image target and background are mainly equivalent, the new segmentation method based on divergence of Itakura- Saito is proposed. Experimental results show that the new thresholding method of this paper is feasible, and the segmentation vision quality of some images by means of thresholding method based on divergence of Itakura- Saito is better than that of Otsu' s method.
出处
《西安邮电学院学报》
2007年第1期80-83,共4页
Journal of Xi'an Institute of Posts and Telecommunications
基金
陕西省教育厅项目(编号:06JK194)