基于神经网络的舌体胖瘦分析  

Analysis plumpness and slenderness of tongue based on the neural net

作　　者：张康[1] 张鹤林[1] 金石[1] 杜建强[1] 丁成华[1] 汪灿华[1] 朱明峰[1] 

机构地区：[1]江西中医药大学,南昌330004

出　　处：《中华中医药杂志》2014年第10期3111-3114,共4页China Journal of Traditional Chinese Medicine and Pharmacy

基　　金：国家自然科学基金项目(No.81260527);江西省卫生厅中医药科研课题(No.2013A085);江西中医药大学校级课题(No.2013ZR0075;No.07375104;No.51203206);江西省自然科学基金(No.20114BAB201030);江西省科技计划项目(No.2011ZBBG70050)~~

摘　　要：舌诊是中医四诊中的重要内容。传统的舌诊方法主观性强,缺乏客观的评价依据,制约了中医舌诊进一步的应用和发展。本研究首先利用计算机技术将舌体从脸部区域提取出来,然后使用二次函数、单峰高斯函数、双峰高斯函数、舍去奇数次项四次函数和含奇数次项四次函数作为拟合函数,对舌体边缘进行曲线拟合,并比较了不同拟合函数对舌体边缘拟合的准确率。最后通过神经网络对舌图像进行胖舌,瘦舌和正常舌的自动归类。Tongue inspection is the main content of TCM four diagnostic methods. Traditional tongue diagnosis has more subjectivity and less objectivity, which hinders further development. This paper made use of computer technology for automatic tongue image segmentation, then used a quadratic function, single-peak Gaussian function and double-peak Gaussian function, rounding an odd number of four times function, contained an odd number of four times functions as a fitting function, and compared the accuracy of the tongue edge of curve fitting. Finally, it classified the fat, thin and normal tongue by using the neural network.

关 键 词：图像分割 舌体胖瘦 曲线拟合 神经网络 

分 类 号：R241[医药卫生—中医诊断学]