This paper investigates variational region-level criterion for supervised and unsupervised texture-based imagesegmentation. The focus is given to the demonstration of the effectiveness and robustness of this region-basedformulation compared to most common variational approaches. The main contributions of this global criterion are twofold. First, the proposed methods circumvent a major problem related to classical texture based segmentationapproaches. Existing methods, even if they use different and various texture features, are mainly stated as the optimization of a criterion evaluating punctual pixel likelihoods or similarity measure computed within a local neighborhood. These approaches require sufficient dissimilarity between the considered texture features. An additional limitation is the choice of the neighborhood size and shape. These two parameters and especially the neighborhood size significantly influence the classification performances: the neighborhood must be large enough to capture texture structures and small enough to guarantee segmentation accuracy. These parameters are often set experimentally. These limitations are mitigated with the proposed variational methods stated at theregion-level. It resorts to an energy criterion defined on image where regions are characterized by nonparametric distributions of their responses to a set of filters. In the supervised case, the segmentation algorithm consists in the minimization of a similarity measure between region-level statistics and texture prototypes and a boundarybased functional that imposes smoothness and regularity on region boundaries. In the unsupervised case, the data-driven term involves the maximization of the dissimilarity between regions. The proposed similarity measure is generic and permits optimally fusing various types of texture features. It is defined as a weighted sum of Kullback-Leibler divergences between feature distributions. The optimization of the proposed variational crite- - ria is carried out using a level-set formulation. The effectiveness and the robustness of this formulation atregion-level, compared to classical active contour methods, are evaluated for various Brodatz and natural images.

Did you like this research project?

To get this research project Guidelines, Training and Code... Click Here

PROJECT TITLE : A Novel Retinex-Based Fractional-Order Variational Model for Images With Severely Low Light ABSTRACT: A new fractional-order variational model based on Retinex is proposed in this research for low-light photos.
PROJECT TITLE : Variational Bayesian Blind Color Deconvolution of Histopathological Images ABSTRACT: In most whole-slide histology images, two or more chemical dyes are used. In digital pathology, slide stain separation or colour
PROJECT TITLE : Variational Osmosis for Non-Linear Image Fusion ABSTRACT: Non-linear image fusion can be improved by using a novel variational model proposed by us. Osmosis energy terms that are similar to those explored in Vogel
PROJECT TITLE : A Robust Group-Sparse Representation Variational Method With Applications to Face Recognition ABSTRACT: For face recognition applications, we offer a Group-Sparse Representation-based technique (GSR-FR). A non-convex
PROJECT TITLE : Structure-Texture Image Decomposition Using Deep Variational Priors ABSTRACT: Structure-texture image decomposition invariably requires structure pictures to have modest norms in some functional spaces and to share

Ready to Complete Your Academic MTech Project Work In Affordable Price ?

Project Enquiry