Synthetic Handwritten Word Alteration Dataset

The dataset contains two types of word images, namely genuine and altered. Word alteration samples have been created synthetically based on the assumption that alteration of any word in a particular document is made by a different person using a different pen with similar colour ink. This dataset is created using ten blue and ten black pens. A total of 4387 genuine word images are created synthetically for each pen. Similarly, an equal number of altered word images are created for each pen. Few sample word images from the dataset are presented below.

N.B.: The data is freely available for research purpose and can be obtained on request (p.dansena23@gmail.com).