Many variational models for image segmentation can be seen in the literature and most of them uses image data fidelity term which involves a given image information only. Usually an energy functional for such models consist two energy functionals namely, external energy and internal energy. External energy helps to attracts the active contour towards the true boundaries of an object, whereas internal energy help to maintain the smoothness of the dynamic curve. In region based models of image segmentation, the external energy is usually the data fidelity term. In this paper we proposed a new variational model of image selective segmentation which utilizes given image information as well as information from an enhanced version of the given image. Experimental results on some real and synthetic images validate the efficiency and robustness of our proposed model in a very few iteration.