Adaptive Segmentation Model for Images with Intensity Inhomogeneity Based on Local Neighborhood Contrast

Abstract

"

Segmentation of images with intensity inhomogeneity is a significant task\nin the field of image processing, especially in medical image processing and analysis.\nSome local region-based models work well on handling intensity inhomogeneity, but\nthey are always sensitive to contour initialization and high noise. In this paper, we\npresent an adaptive segmentation model for images with intensity inhomogeneity in\nthe form of partial differential equation. Firstly, a global intensity fitting term and a\nlocal intensity fitting term are constructed by employing the global and local image\ninformation, respectively. Secondly, a tradeoff function is defined to adjust adaptively\nthe weight between two fitting terms, which is based on the neighborhood contrast of\nimage pixel. Finally, a weighted regularization term related to local entropy is used to\nensure the smoothness of evolution curve. Meanwhile, a distance regularization term\nis added for stable level set evolution. Experimental results show that the proposed\nmodel without initial contour can segment inhomogeneous images stably and effectively, which thereby avoiding the influence of contour initialization on segmentation\nresults. Besides, the proposed model works better on noise images comparing with\ntwo relevant segmentation models.<\/p>"