An analysis of the transition proportion for binarization in handwritten historical documents
作者：
Highlights：
• The minimum-error-rate (MER) based on transition sets is similar to the actual MER.
• We show that the transition proportion has a small influence in the normal threshold.
• The transition proportion is a feature to calculate the stroke width on a document.
• The transition proportion is a feature that identify regions of interest.
• Gray-intensity histograms are lognormally distributed rather than normally distributed.
摘要
Highlights•The minimum-error-rate (MER) based on transition sets is similar to the actual MER.•We show that the transition proportion has a small influence in the normal threshold.•The transition proportion is a feature to calculate the stroke width on a document.•The transition proportion is a feature that identify regions of interest.•Gray-intensity histograms are lognormally distributed rather than normally distributed.
论文关键词：Historical documents,Threshold,Thresholding,Binarization,Transition method,Transition pixel,Normal,Lognormal,Minimum error rate,Otsu,Bayes׳ theory
论文评审过程：Received 13 September 2012, Revised 27 January 2014, Accepted 7 February 2014, Available online 18 February 2014.
论文官网地址：https://doi.org/10.1016/j.patcog.2014.02.003
