Skeleton Hinge Distribution for Writer Identification

Abstract

In this paper, a feature that is based on statistical directional features is presented. Specifically, an improvement of the statistical feature: edge hinge distribution, is attempted. Furthermore, different matching techniques are applied. For the evaluation, the Firemaker DB was used, which consists of samples from 250 writers, including 4 pages per writer. The suggested feature, the skeleton hinge distribution, achieved accuracy of 90.8% using nearest neighbor with Manhattan distance for matching.

Directional Hinge Features for Writer Identification: The Importance of the Skeleton and the Effects of Character Size and Pixel Intensity

“…In our previous work at [3], the use of skeleton information in the document image instead of the edge information was suggested to improve the previous techniques further. As a result, the Skeleton Hinge Distribution achieved 90.8% accuracy on the Firemaker Data set [5].…”

“…In our previous work on Skeleton Hinge Distribution [3], the skeleton information made the feature extraction faster. However, by considering only the Skeleton, a big part of the available pixel information is discarded.…”

“…One of the main assumptions made in the Skeleton Hinge Distribution [3] is that all stroke widths, i.e., line thickness, should be the same size. Practically this means that an attempt is made to condense all the writing information in 1-pixel width strokes.…”

“…The work presented here is influenced by the Hinge Distribution family of features first described in Edge-direction distribution and Edge-Hinge Distribution [4] and improved in Edge-Hinge Combinations [9]. Later contributions include the Quill-Hinge [11], the Quad-Hinge and Co-Hinge [6] and Skeleton Hinge distribution [3].…”

“…The work presented here entails a statistical analysis of features and is mainly influenced by the Skeleton Hinge Distribution Feature [3]. More specifically, four new Hinge Distribution techniques are presented.…”