Novel level set model for the image segmentation based on parzen-window

The Chan-Vese level-set model reported by Chan-Vese for image segmentation has extensive applications in many fields, but also has some disadvantages such as the weak anti-noise. In this paper, a novel level-set model based on the parzen-window is presented. The evolution of the active contour in our level-set model neither depends on the gradient of the image nor the averages inside and outside the objects, as in the classical level-set models. However, our novel model depends on the parzen-window probability distribution associated with the gray of the image to evolve the active contour. And thus our model becomes less sensitive to the noise of the image. In addition, another algorithm is presented to improve the computational efficiency of the parzen-window estimation. We present the experimental results not only in the application of synthetic image segmentation, but also in the difficult application of synthetic aperture radar (SAR) image segmentation; it is shown that we propose a novel method for image segmentation.