Ancient Handwritings Decomposition Into Graphemes and Codebook Generation Based on Graph Coloring

Abstract

We present in this paper a new method of analysis and decomposition of handwritten documents into glyphs (graphemes) and their associated code book. The different techniques that are involved in this paper are inspired by image processing methods in a large sense and mathematical models implying graph coloring. Our approaches provide firstly a rapid and detailed characterization of handwritten shapes based on dynamic tracking of the handwriting (curvature, thickness, direction, etc.) and also a very efficient analysis method for the categorization of basic shapes (graphemes). The tools that we have produced enable paleographers to study quickly and more accurately a large volume of manuscripts and to extract a large number of characteristics that are specific to an individual or an era.

Unsupervised categorization method of graphemes on handwritten manuscripts: application to style recognition

“…In [20] the codebook was constructed of fragments of strokes by sliding an n x n around the skeleton of a letter and where no rules of decomposition were formulated such as the crossing points, junction points and local minimum points that are presented in [2]. Also many previous methods use one Kmeans method [13] for the construction of a codebook.…”

“…59 features are grouped in to 11 descriptors as follows: {D 1 =Height, D 2 =Width, D 3 =Eccentricity, D 4 =Global density, D 5 = Direction, D 6 =Perimeter, D 7 =Fitness, D 8 =Compactness, D 9 = nine densities, D 10 = eight orientations, D 11 =Zernike Moments} [2]. This will yield 2 59 possible combinations for the features.…”

“…The handwritings decomposition into graphemes takes place at the following points: crossover point, junction and local minimum point of a stroke width (figure 5). The details of the decomposition are presented in [2]. generic weights is equal to 1.…”

“…The decomposition method of the manuscript into graphemes is described in detail in [2]. Each grapheme is later on represented by a set of features.…”

“…This type of analysis can provide a larger point of view on the most banal manuscript revealing a rich source of information for the paleographer. This paper presents a new codebook construction mechanism based on graphemes that are issued from the decomposition of manuscripts that we presented in [2]. This categorization of the shapes of grapheme represents a basis for the computerization of the analysis, transcription and recognition of style or writer.…”