A Novel Image Segmentation Algorithm Based on Improved Active Contour Model

摘要

During executing the image segmentation algorithm based on classical geometric active contour model, to obtain accurate segmentation results always involves a redundantly iterative process. And what's more, sometimes this tedious iteration does not make the algorithm converge on the desired edge and even brings out some overshoot. To improve the segmentation efficiency and accuracy, a novel image segmentation algorithm was presented. First, the gradient image is calculated out based on the vector-valued image and then an adaptive edge indicator is proposed. Second, the revised active contour evolution model using variational level set method is put forward. The experiments demonstrate that the model has significantly increased the convergence rate and accuracy. And the proposed segmentation algorithm has also greatly improved the flexibility of the control of active contour evolution by means of its adaptive parameters adjustment.