基于动态阈值和修正模型的舌体提取算法

Tongue Extraction Algorithm Based on Dynamic Thresholds and Correction Model

针对在舌象中较难准确分离舌体的问题,提出从舌体的颜色、形状、位置等先验知识出发,采用HIS色彩模型去除舌象中的嘴唇、脸部等信息,用动态阈值分割算法提取舌体初始轮廓,用舌体修正模型得到最终舌体。通过舌体提取实验和数据分析表明,该方法不仅在抗噪能力、提取精度方面好于其它算子,而且克服了嘴唇、脸部的影响,对舌体凹陷区域也取得了较好的分割效果。舌体的准确提取为下一步舌质舌苔的准确分离打下了良好基础。

To address the problem of segmenting the tongue body from a tongue picture, a novel algorithm is proposed, which is based on prior knowledge of color, shape and position of tongue body. Using HIS color model, the algorithm preserves the contour of tongue body while information of lip and face tissues is removed accurately. After that, the tongue body contour is refined by dynamic threshold segmentation algorithm. Finally, the tongue body is extracted by a regularization model. Experiments and data analysis show that the proposed method in the ability to resist noise and extract the precision better than other operator. In addi-tion, the method works well in presence of lip and face tissues. good results are obtained by the proposed algorithm in spite of the sunken area of tongue body. Accurate segmentation of tongue body plays an important role of preparing good extraction of coating on the tongue.