基于超光谱图像的舌体分割算法  

AUTOMATED TONGUE SEGMENTATION ALGORITHM BASED ON HYPERSPECTRAL IMAGE

作　　者：李庆利[1] 薛永祺[2] 王建宇[2] 岳小强[3] 

机构地区：[1]华东师范大学信息科学技术学院,上海200030 [2]中国科学院上海技术物理研究所,上海200083 [3]第二军医大学附属长海医院中医科,上海200433

出　　处：《红外与毫米波学报》2007年第1期77-80,共4页Journal of Infrared and Millimeter Waves

基　　金：国家863高技术计划资助项目(2001AA131090)

摘　　要：舌体分割是实现中医舌诊现代化的一个前提工作,分割质量直接影响舌象诊断的性能.然而由于舌的生理特征比较复杂,舌的形态以及舌与嘴唇的连接各不相同,使得舌体分割比较困难.本文采用超光谱成像系统代替数码相机采集舌图像,并借鉴光谱角度匹配算法,发展了基于超光谱图像的舌体分割算法.这一算法通过光谱角度匹配算法将超光谱舌图像立方体转换成光谱角度立方体,然后采用一维脉冲波形的边沿信息检测出边缘信息,最后实现了舌体的分割.试验表明,这种基于超光谱图像的舌体分割算法可以较为准确地将舌体分割出来.The automated segmentation of the tongue body is a premise to establish an automatic diagnosis system according to the features of tongue in traditional Chinese medicine （ TCM ）, whose qualities have great effect on the performance of tongue diagnosis. However, automated tongue segmentation is difficult clue to the complexity of pathological tongue, variance of tongue shape and interference of the lips. Here a novel algorithm for automated tongue segmentation was presented based on hyperspectral tongue image data acquired from a hyperspectral imaging system. First, by finding the spectral angle （SA） between each pixel and every other pixel in the original data cube, a transformed data cube was constructed. Thus, each spectrum in the transformed SA cube contained information about spatial changes in the tongue scene. Then, each spectrum in the transformed SA cube was analyzed with a one - dimensional edge - detector. Finally, the whole contour of the tongue was extracted from the hyperspectral tongue image according to the edge detected. Experimental results demonstrate that the novel tongue segmentation algorithm can segment the tongue more accurately.

关 键 词：中医 舌体分割 超光谱 舌诊 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]