A Characteristic Function-Based Algorithm for Geodesic Active Contours

Active contour models have been widely used in image segmentation, and the level set method (LSM) is most popular approach for solving models, via implicitly representing by a function. However, LSM suffers from high computational burden numerical instability, requiring additional regularization terms or re-initialization techniques. In this paper, we use characteristic functions to represent contours, propose new representation geodesic active contours derive an efficient algorithm termed as iterative convolution-thresholding (ICTM). Compared LSM, ICTM simpler much more efficient. addition, enjoys desired features of set-based methods. Extensive experiments, on 2D synthetic, ultrasound, 3D CT, MR images nodule, organ lesion demonstrate that proposed not only obtains comparable even better segmentation results (compared LSM) but also achieves significant acceleration.