改进的GrabCut方法在舌诊系统中的应用

数字舌象图片的正确分割是实现计算机中医舌诊自动化系统的重要前提,为了高效而准确地分割出舌象,提出了一种基于四叉树与GrabCut的舌象分割方法.该方法首先利用四叉树分解对采集的舌象初分割,然后用相似区域的颜色均值优化GrabCut算法中高斯混合模型参数,最终完成舌象分割.实验结果表明：该算法使得舌象分割效率大大提高,具有很好的实用性.

Application of improved GrabCut method in tongue diagnosis system

Correct segmentation of digital tongue is an important prophase work to establish tongue diagnosis automatic system of traditional Chinese medicine （TCM）, in order to divide tongue image efficiently and accurately, present a segmentation method based on quad-tree and GrabCut. The method uses quad-tree decomposition to divide collected tongue image first, and then uses mean color of similar regional, optimize Gaussian mixture model （GMM） parameters of GrabCut algorithm, finally achieve tongue image segmentation. Experimental results show that this method increases efficiency of tongue image segmentation greatly, and has a good practicability.