A hybrid active contour image segmentation model with robust to initial contour position

In the present study, a hybrid active contour image segmentation model was proposed, which combines the local and global information of an image. Said model can be used to prevent the problem that the evolution curve is easy to fall into local optimum when the active contour model uses the local information of the image to segment the image with intensity inhomogeneity. In the model, firstly, the local energy term is constructed by using the image intensity mean of the local region inside and outside the evolution curve to capture the intensity inhomogeneity of the image. Secondly, the global energy term is constructed by using the intensity mean inside and outside the evolution curve to drive the evolution curve to the target edge. Finally, to adaptively adjust the relationship between the local energy term and the global energy term, the weight coefficient is constructed by the gray level of the local and global regions of the image. As such, the proposed model can adaptively adjust the evolution of the curve with the change of the target region. Experimental results on natural images and brain tumor images show that compared with the traditional and several of the latest active contour models, the proposed model has higher segmentation accuracy and robustness to the initial contour.