Recovery of writing sequence of static images of handwriting using UWM

It is generally agreed that an on-line recognitionsystem is always reliable than an off-line one. It is due tothe availability of the dynamic information, especially thewriting sequence of the strokes. This paper presents anew statistical method to reconstruct the writing order ofa handwritten script from a two-dimensional static image.The reconstruction process consists of two phases, namedthe training phase and the testing phase. In the trainingphase, the writing order with other attributes, such aslength and direction, are extracted from a set of trainingon-line handwritten scripts statistically to form auniversal writing model (UWM). In the testing phase,UWM is applied to reconstruct the drawing order of off-linehandwritten scripts by finding the highest totalprobability. 300 off-line signatures with ground truth areused for evaluation. Experimental results show that thereconstructed writing sequence using UWM is close to theactual writing sequence.