An Efficient Approach for Handwritten Devanagari Script Recognition

EasyChair Preprint 4370

Abstract

Handwritten text recognition is a challenging task because of the vast changes in writing styles. In India, a massive number of people use Devanagari Script to write their documents, but due to large complexity, research work accomplished on this script is much lesser as com­pared to English script. Hence, recognition of handwritten Devanagari Script is amongst the most demanding research areas in the field of image processing.  Feature extraction and recognition are key steps of OCR which affects the accuracy of the character recognition system. This paper gives an efficient approach for handwritten Devanagari script recognition based on various parameters like database used, sample size, training and test set ratio, class size, data normalization size, recognition accuracy, etc.

Keyphrases: ANN, CNN, DBN, DCNN, Devanagari script, KNN, OCR, SVM

@booklet{EasyChair:4370,
  author    = {Manoj Sonkusare and Roopam Gupta and Asmita Moghe},
  title     = {An Efficient Approach for Handwritten Devanagari Script Recognition},
  howpublished = {EasyChair Preprint 4370},
  year      = {EasyChair, 2020}}