This thesis proposes a drawing and writing tool recognition algorithm based on features calculated from the shape of stroke endings.
The analysis of strokes is an interdisciplinary field which unites different fields of research: Art History, Paleography and Computer Science, especially the field of Image Processing and Pattern Recognition. The application for this method is to help art historians to identify the drawing tool used for a drawing. Since the style of a drawing depends on the drawing tool used, drawing tool recognition is an important step toward a style analysis. A stroke is a fundamental part of drawing and writing. This means that every drawing and every handwritten document consists of an assemblage of strokes which can be arranged in different ways.
Before the endings can be extracted, the strokes need to be separated from the background.
This results in a binary image which is then used to extract static open and half-open strokes. A static stroke is de ned as a stroke which is delimited by two endpoints, two knots or an endpoint and a knot, where a knot is the junction of at least three strokes. If the stroke connects two endpoints, it is called open, and if it is delimited only by one endpoint, it is called half-open.
This leads to the conclusion that endings can be extracted from open and half-open static strokes. They are extracted according to their varying width in contrast to the remainder of the stroke. A stroke formation is partitioned into static strokes with the help of a skeleton. Several features regarding curvature, proportions etc. are calculated out of the shape of the endings. These features are then used to classify stoke endings with a Support Vector Machine Classifier. Test data consists of synthetic strokes, strokes and stroke formations drawn with different drawing and writing tools and glagolitic characters from an ancient missal from St. Catherines Monastery and made by a calligrapher.