Handwriting Assessment Using a Haptic Joystick for Rehabilitation Purposes

Abstract

Writing skills play a major part in communication, whether in a typing or in a handwriting scenario. Thus, a decrease or even loss of these skills can be a major setback. The system described herein aims to promote handwriting rehabilitation through haptic guidance and analysis of characters written using a haptic joystick. The first goal of this work is the creation of a system that encourages a user to write better through repetition, and thus have an impact on the proprioceptive system. In the second part of this work, there is a character analysis model evolution. Character analysis is performed by digit or letter classification and quantification of written characters quality. As such, different methods are studied and evaluated to achieve the best classification accuracy. Regarding handwritten characters classification, the pre-eminent procedure for digits consists in the use of Histogram of Oriented Gradients coupled with a multiclass Support Vector Machine (HOG-SVM) whereas for letters the strategy involves using a Convolutional Neural Network (CNN). For handwriting quality quantification, Dynamic Time Warping (DTW) was performed between the written character and a reference image of the same character. Finally, haptic guidance impact in a handwriting retraining in rehabilitation scenario is evaluated.

Keywords

• CNN
• DTW
• handwriting rehabilitation
• haptic joystick
• HOG-SVM