R-HOG Feature-Based Off-Line Odia Handwritten Character Recognition

R-HOG Feature-Based Off-Line Odia Handwritten Character Recognition

Abhisek Sethy (Koneru Lakshmaiah Education Foundation, India) and Prashanta Kumar Patra (College of Engineering and Technology Bhubaneswar, India)
Copyright: © 2020 |Pages: 15
DOI: 10.4018/978-1-7998-0066-8.ch010

Abstract

Offline handwritten recognition system for Odia characters has received attention in the last few years. Although the recent research showed that there has been lots of work reported in different language, there is limited research carried out in Odia character recognition. Most of Odia characters are round in nature, similar in orientation and size also, which increases the ambiguity among characters. This chapter has harnessed the rectangle histogram-oriented gradient (R-HOG) for feature extraction method along with the principal component analysis. This gradient-based approach has been able to produce relevant features of individual ones in to the proposed model and helps to achieve high recognition rate. After certain simulations, the respective analysis of classifier shows that SVM performed better than quadratic. Among them, SVM produces with 98.8% and QC produces 96.8%, respectively, as recognition rate. In addition to it, the authors have also performed the 10-fold cross-validation to make the system more robust.
Chapter Preview
Top

Introduction

Automatic Character Recognition has considered as to be one of the emerging fields of Pattern Recognition. Henceforth it was quite impressive for researchers to do more qualitative work to solve the real-world problems. An automated recognition model has been developed not only for printed, handwritten characters along with degrade characters are also reported by Govindan et al. (1990). Each recognition system has certain intermediate stages such as acquisition, preprocessing, feature extraction and last the classification are accepted in well- defined manner so to report high recognition rate (Mantas, 1986). As on context of recognition of handwritten one is quite challenging as compared with printed one, it is so because handwritten more complex in nature and various individual’s various way of writing skills in terms of shape orientation of writing (Plamondon et al., 2000). In this paper we have try to build a recognition system for Odia Handwritten characters. Handwritten Character of Odia is one of the oldest scripts and the official language of Odisha state, India. In this regard several recognition techniques have been evolved for variance kind of languages but writing pattern of Odia character is just like as curve appearance. Hence, it adds more complex analysis to the recognition model for handwritten ones (Pal et al., 2004).

In context to automatic recognition handwritten Odia Characters is most challenging task among the individual’s characters and its due that most the characters are mostly alike to each other that is in terms of shapes size and orientation. These reported variations make researchers to work more to provide robust techniques towards this problem domain. To have a successive recognition accuracy for any OCR system one should put more emphasis on feature extraction technique. There was certain well-defined feature extraction technique has been listed up which was used in recent year is described by to have a proper solution to above mentioned (Pal et al., 2012; Sethi et al., 1977). Here in this paper we have shown how cell-based approach is quite efficient to achieve high recognition rate for the hand-written characters. For implementing a cell based we have harnessed the Rectangle Histogram Oriented Gradient (R-HOG) over the data image in the feature extraction segment. In addition to it we have shown the reduction in feature dimension through implementing Principal Component Analysis (PCA) (Sethy et al., 2018). At last the reduced feature vector are forwarded to SVM and Quadratic Classifier and report the respective recognition rate. In order to make the recognition system effective we have segmented the model into certain section. Where section 2 represents the related work done by various researchers, section 3 depicts the overall proposed model where all the significant stages are explained precisely, and Section 4 and Section 5 represents the result analysis and conclusion along with future scope of the proposed model.

Complete Chapter List

Search this Book:
Reset