摘要
舌像分割是数字化舌诊的第一个关键步骤。为了准确地从图像中分割出舌体,提出了结合亮度信息和粗糙度信息的分割方法。首先使用最大类间方差法在舌图像的亮度范围内自动选取阈值,提取出较亮的舌体部分,然后再利用粗糙度信息进一步剔除非舌体部分,最后使用数学形态学的膨胀、腐蚀和开运算操作等方法进行后续处理,得到最终结果,成功地实现了舌像分割的功能。实验结果表明,该方法具有较好的效果。
Tongue image segmentation is the first important step of tongue diagnosis digitization. To segment tongue from image efficiently, a method based on luminance and roughness was proposed. First the Otsu method was implemented to select a threshold in luminance field of the image automatically. Then, non-tongue regions were eliminated with the difference between the roughness of tongue and skin. Finally, ultimate result was produced by employment of erosion, dilation and opening operation in mathematical morphology. The experimental result shows that this method is significant and efficient.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2006年第z1期374-376,379,共4页
Journal of System Simulation
基金
国家自然科学基金委创新研究群体基金(60521002)
国家自然科学基金(60573147)。
关键词
舌像分割
亮度
粗糙度
最大类间方差法
tongue image segmentation
luminance
roughness
Otsu method