Devanagari Script Recognition System using Combined Approach

Abstract

We envisioned serving simple technique with novel idea for the newcomers, working in the Devnagari Optical Character Recognition (DOCR) area. In this system, template matching technique and statistical technique is combinely used for the character recognition. Statistical information is considered as a feature of an image and here pixels values are considered as feature to achieve the recognition. The numbers of comparisons are drastically reduced compared to traditional template matching technique hence performance of the system is improved. The correlation and related methods are used to find images that are similar to a template. The results are compared with results of template matching technique in terms of elapsed time that is time required to recognize the character here. The recognized devanagari characters then translated in to English through Google translator using Matlab which will provide human assistance in different applications. In English OCR results were of such type that’s it can produce technology driven applications. The algorithm used is very effective and can serve as a basis for further research towards other similar Indian scripts.

Keywords: Correlation, devanagari optical character recognition, segmentation,statistical technique, template matching