Handwritten Word Recognition of Various Languages: A Review

Handwriting recognition (HWR) is the ability of a computer to interpret and understand handwritten text. HWR involves gathering, storing, and processing digital pen data. HWR is used for many tasks, including writing automation, digital document archiving and retrieval, text entry automation, and document data extraction. Handwriting recognition software is used to help automate these tasks or to make them easier for the user. HWR is also used to improve accessibility for people with disabilities who cannot generate characters traditionally accurately. Recognizing the handwritten word is a complicated process because of the nature of the characters, such as similar shapes and sizes, spaces between the characters, discontinued characters, overlapped characters, complex stroke patterns, and writing styles in various languages. A deep learning approach has been mainly used in handwritten recognition recently. Deep learning can be used to process both labeled and unlabeled data. The advantage of the deep learning approach is to perform feature extraction independently. The survey focuses on the issues and challenges previous research has encountered. The study lists the types of methodologies and datasets utilized in different languages are described.