Authors : G. Roshan, M. Akhil Kumar, R.S.V. Vishnu and G. Virajit
Abstract: Handwritten character recognition has been one of the most active and challenging areas of research in the field of image processing and pattern recognition. This study is a brief survey and comparison of two different feature extraction methods used in recognition of uppercase English alphabets in a given handwritten scanned text from sentences. These techniques were tested using feed-forward neural networks which are trained using back propagation algorithm with the cost function as Mean Square Error (MSE). The resesrch also provides substantial evidence that enhancement of recognition and reduction of misclassification is dependent on the type of features used. We extended our researcch towards extraction of characters from colored and non-uniform backgrounds using Maximally Stable External Regions (MSER) feature algorithm.
G. Roshan, M. Akhil Kumar, R.S.V. Vishnu and G. Virajit, 2017. Feature Extraction Techniques for Handwritten Character Recognition using Neural Networks with Non-Uniform Background. Journal of Engineering and Applied Sciences, 12: 9359-9364.