基于Perona-Malik方法的GVF Snake模型在舌像分割中的应用

Application of GVF Snake model based on Perona-Malik algorithm in segmentation of tongue image

采用Snake模型进行舌像分割虽然能够快速地得到舌体轮廓信息,但是Snake模型对于噪声和外部图像能量十分敏感,特别是舌头颜色与嘴唇以及面部颜色相近,分割时容易丢失重要信息。为了能够有效地将这些信息分离得到准确的舌体轮廓,提出了一种改进的Snake模型。该模型首先引入Perona-Malik方法来处理原始图像的边缘信息,强化模糊轮廓界线,再计算得到舌体区域的梯度能量场,以此作为Snake模型的外部能量部分,得到最终的Snake轮廓曲线。实验表明,使用改进的模型可以得到包含更多边缘信息的舌体图像,特别是颜色相近或者轮廓线不规则的舌体图像效果更加明显。

Using the Snake model on tongue image segmentation can get tongue contour information quickly,but the Snake model is very sensitive to image noise and the external force,especially the color of tongue and lips is similar,also the face,and it's easy to lose important information of tongue during the segmentation. In order to effectively separate these information and obtain the tongue contours accurately,an improved Snake model is proposed. This model firstly uses the Perona-Malik method to deal with the edge information of original image,highlights the fuzzy boundary contour,and then calculates the gradient vector field of tongue area,as the part of the external force of the Snake model,and obtains the Snake contours eventually. The experimental results show that the improved model can be used to obtain the tongue images with more edge information,especially better at the tongue images with similar color or irregular contours.