Improving the Accuracy for Offline Arabic Digit Recognition Using Sliding Window Approach
- PDF / 586,020 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 104 Downloads / 183 Views
RESEARCH PAPER
Improving the Accuracy for Offline Arabic Digit Recognition Using Sliding Window Approach Ebrahim Al-wajih1,2
•
Rozaida Ghazali1
Received: 13 September 2018 / Accepted: 18 January 2020 Shiraz University 2020
Abstract Handwritten Digit recognition is a challenging problem these days due to the widely used Arabic language in the world, especially in the Middle East region. In this paper, sliding windows are used to enhance classification accuracies and implemented using random forests (RF) and support vector machine (SVM) classifiers for recognition of Arabic digit images. In order to study their effectiveness with and without using sliding windows, four different feature extraction techniques have been proposed which includes Mean-based, Gray-Level Co-occurrence Matrix (GLCM), Moment-based, and Edge Direction Histogram (EDH). The obtained accuracies show the significance of using sliding windows for classifying digit. The recognition rates acquired using the modified version of AHDBase dataset are 98% when Meanbased and Moment-based are applied with RF classifier, 98.33% and 99.13% when GLCM and EDH are used with linearkernel SVM, respectively. Moreover, the performance of this study is compared against recent state-of-the-art approaches, namely Geometric-based, two-dimensional discrete cosine transform, Hierarchical features, Hetero-features, Discrete Fourier Transform and geometrical features, Gabor-based, gradient, structural, and concavity and Local Binary Convolutional Neural Networks. Keywords Arabic digit recognition Sliding window Texture features Gray-level co-occurrence matrix Edge direction histogram
1 Introduction Handwritten digit recognition is still a challenging problem although many studies have been carried out to address it. This is due to the fact that the exponential development of the technology, especially with the smart devices, needs noncomplex algorithms to enhance the accuracy of recognition systems. Further, it is difficult to consider that the problem of identifying numbers was solved because of the limited sizes of datasets which is the common issue in & Ebrahim Al-wajih [email protected] Rozaida Ghazali [email protected] 1
Faculty of Computer Science and Information Technology, Universiti Tun Hussein Onn Malaysia, Batu Pahat, 86400 Parit Raja, Johor, Malaysia
2
Society Development & Continuing Education Center, Hodeidah University, Alduraihimi, 3114, Hodeidah, Yemen
pattern recognition area. The number of writers in the available datasets does not cover all variations of writing styles from person to person. To solve this problem, several different techniques have been proposed by researchers to find the appropriate algorithm. Handwritten digit recognition is categorized into two kinds, online and offline, based on the input method to the system. The applications that used online method receive the input by the movement of the pen on a pen-based screen, while the offline applications use the image of a digit that is captured using an inter
Data Loading...
Recommend Documents
Arabic (Indian) digit handwritten recognition using recurrent transfer deep architecture
221 113 1MB
A Robust Approach with Text Analytics for Bengali Digit Recognition Using Machine Learning
160 1 8MB
