An impact of grid based approach in offline handwritten Kannada word recognition

The scanning of paper documents followed by the storage, retrieval, display, and management of the resulting electronic images, is known as document image processing, which is a subfield of Digital Image Processing. The main objective of the document image analysis is to recognize the text and graphics components in the images. Optical Character Recognition [OCR] is the process of converting the image obtained by scanning a text or a document into machine-editable format. OCR has practical potential applications in writer identification, forensic analysis handwriting, health care, legal, banking, postal services, etc. Recently, handwriting recognition is now gain spread lot of importance due to sources such as paper documents, photographs, touch-screens and other devices. In this paper we study the impact of grid based approach in offline handwritten Kannada word recognition. Popular subspace learning method, i.e. Principal Component Analysis is used for better representation of the given input word. The study is experimented on handwritten word comprising of 28 district names of Karnataka state. The experiment suggest grid based approach outperforms the standard global based approach.