A Level Set Approach to Image Segmentation With Intensity Inhomogeneity

摘要: It is often a difficult task to accurately segment images with intensity inhomogeneity, because most of representative algorithms are region-based that depend on homogeneity the interested object. In this paper, we present novel level set method for image segmentation in presence inhomogeneity. The inhomogeneous objects modeled as Gaussian distributions different means and variances which sliding window used map original into another domain, where distribution each object still but better separated. transformed domain can be adaptively estimated by multiplying bias field signal within window. A maximum likelihood energy functional then defined whole region, combines field, function, piecewise constant function approximating true signal. proposed directly applied simultaneous correction 3 7T magnetic resonance images. Extensive evaluation synthetic real-images demonstrate superiority over other algorithms.