Sarangi et. al, Apeejay - Journal of Management Sciences and Technology, 3 (1), October - 2015 ISSN -2347-5005 58 Handwritten Character Recognition: A Review Dr. Pradeepta K. Sarangi Apeejay Institute of Technology, Greater Noida, U.P, India pradeepta_sarangi@yahoo.com Dr. Kiran K. Ravulakollu Sharda University Greater Noida, U.P, India Alok Kumar Singh Project Manager Syntel Inc, USA Abstract: Character recognition is one of the active research area in the recent trend. It belongs to the categories of pattern recognition. This field of character recognition has gained enormous attention due to its wide application in various fields. This area has grown as one of the potential areas for researchers. This paper describes a review of handwritten character recognition with specific focus on feature extraction. This review is based on an extensive study of numerous published literature including articles in journals, proceedings of conferences and Doctoral theses and mainly focused towards the use of various feature extraction techniques in character recognition with special focus on handwritten Odia characters. Keywords: Character Recognition, Feature Extraction, Odia Characters I. INTRODUCTION Character recognition consists of three key steps: pre-processing, feature extraction and classification. Feature extraction is a process to represents the input image in a miniature form keeping the properties of original image intact. In pattern recognition, feature extraction is considered as a form of dimensionality reduction. This work gains contribution from various individual publications and some of the review papers [1-9] published online and collected form various authors. II. REVIEW OF LITERATURE ON NON-ODIA CHARACTERS Handwritten Character Recognition (HCR), a special form of pattern recognition that deals with automatic recognition of handwritten characters in document form. These could be separated into two categories: printed character recognition and handwritten character recognition. Printed characters are uniform and distinctive where as handwritten characters are non-uniform and mostly depend on the writing style of the author. Handwriting of same writer might vary depending on the situation and material used for writing. It has been a tedious task to develop a system suitable for all types of characters with great accuracy. Hence researchers argue the need of a strong generalized method to identify distorted patterns. According to Deepa et al. [10], character recognition systems incorporate four major phases. Pre- processing of the input images, segmentation, extraction of feature vectors and classification. The authors have recognized handwritten characters through multi-set of puzzle pieces and four stroke features such as horizontal , vertical , left slant and right slant for classification. They conclude that, in pattern