基于总变分方法去噪的测地线活动轮廓的应用

传统的测地线活动轮廓（geodesic active contour）在目标轮廓的提取中虽然能使一条初始曲线朝着目标边界逼近,但是高斯平滑的各向同性性使得图像的边缘信息模糊甚至丢失,这样,曲线的演化过程会变得不稳定而导致演化速度变慢。据此,提出一种基于总变分方法的测地线活动轮廓模型。由于总变分方法可以在去除噪声的同时,对边缘的信息进行增强,实验证明该模型不但能够使曲线精准地收敛到期望轮廓上,且运算时间短。

Application of geodesic active contours based on total variation denoising

Although the classical geodesic active contour can deform an initial curve towards the boundary of the object to be detected, but the Gaussian regularizafion that has the property of the isotropic diffusion may simultaneously blur the edge or boundry of the objects and make the curve propagating unstable which can lead to a time - consuming propagation. This paper introduced a geodesic active contour model based on total variation denoising to solve the problem mentioned above. Total variation model can smooth out the noises in image, but it may simultaneously preserve and sharp the edge of the objects. Experimental results of applying the proposed method in noising images show that it yields good effect and saves time.