基于GVF Snake和边界跟踪的主动轮廓图像分割

Image segmentation based on GVF snake and boundary tracking

梯度向量流蛇(GVF Snake)模型在处理图像分割问题上取得了较好的结果,但它对初始轮廓曲线的依赖程度较大且梯度向量场计算时间较长,故此提出一种基于GVF Snake模型和边界跟踪的轮廓提取图像分割算法。该算法利用边界跟踪算法进行粗糙的分割,获取边缘位置有效信息点,经采样后生成一条初始轮廓线。同时,基于拉格朗日法求解梯度向量场的方法,提出一个距离终止条件以提高计算速度。实验结果表明,与GVF Snake、手动GVF Snake和CV活动轮廓算法相比,该算法有效提高了图像分割的自动化程度和分割精度。

Gradient vector flow active contour method in dealing image segmentation have achieved good results,but it largely depends on the initial contour curve and the long time cost of compute the gradient vector field. A method which is based on GVF Snake model for image segmentation is proposed. The method uses the boundary extraction algorithm for coarse segmentation,to obtain effective information points edges postion,then samples the information points to generate the initial contour. Meanwhile,the Lagrangian method is used to numerically compute the gradient vector field,a termination condition based on distance is given. which can effectively improve the computing speed. Experiments performed on standard test images showed that the method,compared with the traditional,manual method and CV active contour,which effectively improves the degree of automation and guarantees relatively segmentation precision in image segmentation processing.