An adaptive Geodesic Active Contour model
Bo Zhang, Yongli Su, Yongfeng Xu, Shuling Zhang · 2010 Sixth International Conference on Natural Computation · 2010
In order to solve the shortcoming of Geodesic Active Contour model that would possibly sink into the non-ideal local minimum when segmenting the objects having concave boundary, an adaptive Geodesic Active Contour model was presented. The new model could adjust the evolution speed of curve based on the curvature of curve and gradient of image by adding an acceleration item in the original model. Moreover, in order to eliminate the influence of noise or false edge, a new pre-process Sobel operator combined the Gaussian filter and calculation of gradient of Geodesic Active Contour model is presented. The results of experimental comparison show that the Sobel operator could simultaneously smooth the image and calculate the gradient, and increase the speed of algorithm, while the new model could segment the image accurately.