Development of potential dysgraphia handwriting dataset

Abstract

This report presents a dataset of offline handwriting samples among Malaysian schoolchildren with potential dysgraphia. The images contained Malay sentences written by primary school students and children under intervention by the Malaysia Dyslexia Association (PDM). Students were expected to copy and write the sentences provided on the paper form that was used to gather data. Students were required to write three sets of sentences. The paper was digitalized by scanning it and converting it into digital form. Furthermore, the images were pre-processed using image processing techniques by converting the images into binary format and interchanging the foreground and background colors. The images were then classified into two categories, namely potential dysgraphia and low potential dysgraphia. The dataset comprised a total of 249 handwriting images, obtained from a sample of 83 participants who were selected in the data collection process, with 114 for potential dysgraphia and 135 for low potential dysgraphia. Both categories of handwriting images were prepared in black and white images.

Keywords

• Data acquisition
• Image preparation
• Handwriting analysis
• Dysgraphia handwriting classification