Image-based historical manuscript dating using contour and stroke fragments. (October 2016)

• Record Type: Journal Article
• Authors: He, Sheng; Samara, Petros; Burgers, Jan; Schomaker, Lambert
• Is Part Of: Pattern recognition. Volume 58(2016:Oct.)
• Page Start: 159
• Page End: 171
• Publication Date: 2016-10
• Subjects: Historical manuscript dating -- Writer identification -- Contour fragment -- Stroke fragment -- Handwriting style; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2016.03.032
• Languages: English
• ISSNs: 0031-3203

Abstract

Abstract: Historical manuscript dating has always been an important challenge for historians but since countless manuscripts have become digitally available recently, the pattern recognition community has started addressing the dating problem as well. In this paper, we present a family of local contour fragments ( k CF) and stroke fragments ( k SF) features and study their application to historical document dating. k CF are formed by a number of k primary contour fragments segmented from the connected component contours of handwritten texts and k SF are formed by a segment of length k of a stroke fragment graph. The k CF and k SF are described by scale and rotation invariant descriptors and encoded into trained codebooks inspired by classical bag of words model. We evaluate our methods on the Medieval Paleographical Scale (MPS) data set and perform dating by writer identification and classification. As far as dating by writer identification is concerned, we arrive at the conclusion that features which perform well for writer identification are not necessarily suitable for historical document dating. Experimental results of dating by classification demonstrate that a combination of k CF and k SF achieves optimal results, with a mean absolute error of 14.9 years when excluding writer duplicates in training and 7.9 years when including writer duplicates in training. Abstract : Highlights: A new image-based historical manuscript dating problem is proposed. We present a family of local contour fragments and stroke fragments features. Historical manuscript dating is performed by writer identification and classification.