A method of neighbor classes based SVM classification for optical printed Chinese character recognition

Area of Science:

• Computer Science
• Artificial Intelligence
• Pattern Recognition

Background:

• Optical Printed Chinese Character Recognition (OPCCR) relies on effective classification methods.
• Support Vector Machines (SVM) are powerful binary classifiers but computationally intensive for multi-class problems like OPCCR.

Purpose of the Study:

• To address the computational inefficiency of standard SVMs in multi-class OPCCR.
• To introduce a novel Neighbor Classes based SVM (NC-SVM) algorithm.

Main Methods:

• Developing and implementing the Neighbor Classes based SVM (NC-SVM) approach.
• Conducting experimental evaluations of NC-SVM for OPCCR tasks.

Main Results:

• The proposed NC-SVM effectively reduces computation time compared to traditional multi-class SVM methods.
• Experimental results demonstrate the efficiency gains of NC-SVM in OPCCR.

Conclusions:

• NC-SVM offers a computationally efficient solution for multi-class OPCCR.
• The method successfully mitigates the time-consuming nature of SVMs in this domain.