An Adaptive Scale Active Contour Model Based on Information Entropy for Image Segmentation

In view of the problem that the fixed scale active contour model cannot quickly and accurately segment images with intensity inhomogeneity, an adaptive scale active contour model based on the information entropy is proposed for image segmentation. Firstly, we put forward a novel energy function by using Maximum Posterior Probability (MAP) and Bayes classification criterion, which greatly improve the ability to extract the image intensity information and the segmentation accuracy for inhomogeneous images. Secondly, we construct an adaptive scale operator by using the image information entropy to let the model can automatically adjust the scale according to the intensity inhomogeneity degree of the image, which improves the segmentation speed of the model. Finally, in order to verify the superiority of our model, we make a comparison between our model and LGDF model, and also make an objective and quantitative analysis of the segmentation results by using the segmentation time, the number of iterations and the similarity index. The final results show that the proposed model not only has high robustness to the initial contour, but also has high accuracy and efficiency in segmenting images with intensity inhomogeneity. © 2017, Editorial Board of Journal of Northwestern Polytechnical University. All right reserved.