Bayesian damage recognition in document images based on a joint global and local homogeneity model

作者：

Highlights：

• A global homogeneity measure is developed by characterizing connected pixels using a probabilistic graph model.

• A local homogeneity measure is developed by modeling neighborhood transition around an individual pixel based on propagation of wavelet approximation coefficients.

• Global and local homogeneity measures are jointly modeled using a Bayesian framework with a Markov random field (MRF) prior.

• Probabilistic recognition of damages in document images based on the joint global and local homogeneity model and Markov chain Monte Carlo (MCMC) sampling.

摘要

•A global homogeneity measure is developed by characterizing connected pixels using a probabilistic graph model.•A local homogeneity measure is developed by modeling neighborhood transition around an individual pixel based on propagation of wavelet approximation coefficients.•Global and local homogeneity measures are jointly modeled using a Bayesian framework with a Markov random field (MRF) prior.•Probabilistic recognition of damages in document images based on the joint global and local homogeneity model and Markov chain Monte Carlo (MCMC) sampling.

论文关键词：Damage recognition,Text homogeneity,Neighborhood transition,Propagation of wavelet approximation,Bayesian inference

论文评审过程：Received 8 December 2020, Revised 8 April 2021, Accepted 10 May 2021, Available online 20 May 2021, Version of Record 4 June 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.108034