Pulse Coupled Neural Network Based Laws' Mask for Devanagari Character Recognition

Manuscripts are the stock of information for forthcoming generations. It is also valuable basis for detecting materials from the Middle Ages. Recognition of manuscript is a challenging task as very few approaches are available to recognize regional languages such as Devanagari, Gurmukhi, Sanskrit, etc. Texture analysis can be considered as the earliest study of human visual perception and utilized for pattern recognition. Laws' Mask is utilized for the extraction of features. The task of image enhancement can be achieved by a new generation of neural networks, particularly, Pulse Coupled Neural Network (PCNN). In this work, Laws' mask is integrated with PCNN to recognize Devanagari characters. The process is assessed using Devanagari Handwritten Character Dataset (DHCD) database. The result illustrate that the suggested approach has achieved better results than the original network.