A Document Image Restoration Algorithm based on Markov Random Field

Ruiguo Yang, Min Zhu, and Youguang Chen


document analysis, image processing and analysis, Markov Random Field (MRF), Bayesian theorem


A new document image restoration algorithm is presented based on the theory of Markov Random Field and the Bayesian theorem. The algorithm consists of two parts: sample training and image restoration. By training samples through the analysis of a large number of specific field’s document images, the probability relationship of image blocks is achieved. Based on such probability, the image restoration conducts the recovery of distorted images. This algorithm has been proven practical and efficient by the simulated experiments.

Important Links:

Go Back