Recognition of similar shaped isolated handwritten Gurumukhi characters using machine learning

Existence of similar shaped handwritten characters in any script affects its automated recognition. Effective recognition of similar shaped characters increases overall performance of automated handwritten character recognition. It is intensive research area because of its application in wide ambit of real human beings problems like postal address interpretation and signature verification in bank cheque processing etc. In this work, people in the age group of 20-50 years were asked to write characters of Gurumukhi Script in their natural writing style. Preprocessing steps were applied on the collected data and zoning based structural & region features were extracted. Data set was created using the extracted features and recognition was done using Random Forest and C4.5 machine learning algorithms.