Segmentation of shadow images using IR based adaptive Level set model

Abstract

Images captured under non uniform lightening and illumination conditions consist of unwanted shadow regions. Segmentation of these images using conventional level set methods show improper results which is a major problem in many real time applications such as medical imaging, pattern recognition etc. In this paper we propose an adaptive Level set model by introducing a new adaptive force function on the level sets evolution based on reflectance & illumination present in the image. This method is proved to be successful in segmentation of images by excluding the shadow regions present in the images.