A Fast Active Contour Model Driven by Global-Local Statistical Energy

Recently, active contour models based on local information have emerged in image segmentation. These models are more robust to local variations of region of interest. But it also brought some new problems, such as local minimum, higher computational cost. To effectively alleviate these problems, this paper presents a novel fast active contour model driven by global-local statistical energy. Firstly, a new local statistical model is constructed to capture the border of object more accurately. Secondly, global image features of image are integrated with local features so as to create an improved model which can avoid local minimum effectively. In addition, an accelerating factor is introduced into global-local statistical model in order to drive contour evolution rapidly. Finally, we carry out experiments on synthetic and real data, and results compared to relevant models have demonstrated robustness, accuracy, and time efficiency of our proposed model.