Stroke Segmentation of Handwritten Text Based on Medial Representation

The paper deals with structural analysis of handwritten text by digital binary images. A new approach to structural analysis is proposed. It consists in representing text image as strokes, a set of calligraphic elements. The method of stroke segmentation of handwritten text is based on the medial representation of a binary image in the form of a continuous skeleton and a radial function. Each stroke is assumed to be a trace of pen movement. The stroke description includes static and dynamic components. The static components of strokes are cycles and chains—subgraphs of the skeleton graph. The dynamic components are the description of strokes as paths of pen movement including the starting point and direction of the movement.