Advancements in Handwritten Text Recognition: A Survey of Methods, Challenges and Future Directions

Handwritten Word Recognition (HWR), also referred as Handwritten Text Recognition (HTR), includes a computer’s ability to acquire and comprehend understandable information written by hand, sourced from materials like paper documents, images, touch screens, and other devices. This technology is a subset of Intelligent Document Processing (IDP), which refers to the use of various technologies, including optical character recognition (OCR), natural language processing (NLP), and machine learning, to extract valuable information from documents. The primary goal of HWR is to interpret individual handwritten words, facilitating the digital conversion of handwritten text. Despite being a subject of study for over four centuries, certain key challenges in handwriting recognition remain unresolved. This survey paper aims to delve into these issues and give better understanding on the provided solutions. Additionally, the paper discusses various approaches in the field, such as offline and online handwriting recognition, providing a comprehensive exploration of methods to tackle the intricacies of handwriting recognition. The ultimate objective of the paper is to elucidate diverse strategies for resolving the challenges associated with handwriting recognition and provide future directions for the improvisations that can be made.