The Use of Dynamic Characteristics in Handwriting Recognition Tasks

In the presented paper different approaches are investigated to recognize handwritten characters represented as points of the pen trajectory. In doing so, datasets in different languages are investigated. In addition, a method for extracting dynamic characteristics from a data set of handwritten characters is described. The peculiarity of the method is that the analysis of motion focuses on the oscillatory component of the writer's hand motion. After preprocessing the trajectory data, a very compact vector of features characterizing the symbol is obtained, which leads to a significant reduction in the amount of resources and time required for processing. The proposed method can significantly improve the accuracy of optical recognition of handwritten characters. Also, the extracted information can be used in other fields related to character recognition, such as analysis of glyphs or signatures, as well as analysis or transfer of writing style.