Color image segmentation and parameter estimation in a markovian framework
Pattern Recognition Letters
IRGS: Image Segmentation Using Edge Penalties and Region Growing
IEEE Transactions on Pattern Analysis and Machine Intelligence
Decoupled Active Contour (DAC) for Boundary Detection
IEEE Transactions on Pattern Analysis and Machine Intelligence
A relationship between the recursive least squares update andhomotopy continuation methods
IEEE Transactions on Signal Processing
Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images
IEEE Transactions on Pattern Analysis and Machine Intelligence
Hi-index | 0.00 |
In the color image segmentation, the usual approach is to segment the image while attempting preserve strong edges. But often in real world, many weak edges or poorly defined edges need to be preserved in order to achieve meaningful segmentation. In this paper, we have proposed a color image segmentation scheme in statistical framework, where the weak edges have been reinforced together with the strong edges. In this regard we have proposed the notion of graduated edge penalty function based bilevel line field to take care of strong as well as ill defined edges in the a priori image modeling. We have also used our previously proposed Double Constrained Compound MRF model as the image prior. The image labels have been estimated using MAP estimation criterion and the model parameters estimation problem has been formulated in Maximum Conditional Pseudolikelihood (MCPL) framework. The Maximum a Posteriori(MAP) estimates of the labels have been obtained by a hybrid algorithm consisting of Simulated Annealing(SA) and Iterated Conditional Mode(ICM) algorithm. The proposed schemes, when compared with Yu 's method exhibited improved performance.