A Multiplication optimization Level Set Algorithm for Image segmentation

Intensity inhomogeneity often occurs in a large number of images, offering a great challenge in image segmentation. The local intensity clustering (LIC) is a region-based method for image segmentation, with the bias field demonstrating good performance in the intensity homogeneity region. However, due to the characteristic of the smoothing and slow change in the bias field, LIC model fails to obtain good results for images with complex background. To solve this problem, this paper proposes a multiplication optimization level set image segmentation algorithm. First, the bias field is represented by a set of multiplication smooth basis functions, with the iterative solution of the bias field converted to search for the optimal coefficients for the basis functions, to reduce the complexity of background. Minimization of this objective function are deduced theoretically to provide the evolution function. Compared with LIC model, our algorithm can simultaneously segment images and correct the bias field. The effectiveness of the algorithm is verified by experiments on synthetic images and medical images.