Markov random field regularisation models for adaptive binarisation of nonuniform images

Two related MRF models, an edge-preserving smoothing model followed by a mo dified standard regularisation, are presented for the adaptive binarisation of nonuniform images in the presence of noise. In particular, a computatio nal model is developed for a modified standard regularisation method which calculates the adaptive threshold surface for noisy images. Since the modif ied standard regularisation depends only on the image data, and not its edg e segments, it gives much better performance and can be applied to more cla sses of image than those methods that solely rely on edge segments. Experim ental results demonstrate that the proposed method has the best performance over three other commonly used adaptive segmentation methods and is faster than previous interpolation-based thresholding techniques.