Open AccessHybrid Feature Extraction and Multiclass SVM Based Ancient Character Recognition
Author(s) -
P. Balasubrmanie*,
E.K. Vellingiriraj
Publication year - 2019
Publication title -
international journal of recent technology and engineering
Language(s) - English
Resource type - Journals
ISSN - 2277-3878
DOI - 10.35940/ijrte.c6520.098319
Subject(s) - artificial intelligence , pattern recognition (psychology) , computer science , character (mathematics) , support vector machine , feature extraction , segmentation , feature (linguistics) , gabor filter , filter (signal processing) , computer vision , mathematics , linguistics , philosophy , geometry
Ancient character recognition is the most difficult task due to their different formats and less frequent knowledge about the ancient characters. This is performed in the previous research work namely Shape and Size aware Character Recognition and Conversion System (SSCR-CS). This research work proved better character recognition outcome. However this research work might be reduced in performance with lesser detailed information in the images. This is resolved in the proposed research work by introducing the most recent techniques for the character recognition outcome. This is attained by introducing the method namely Hybrid Feature Extraction and Multiclass SVM based recognition method (HFE-MCSVM). In this research work, initially image preprocessing is performed by using Gabor filter. After preprocessing segmentation of characters is performed by using overlapped character segmentation method. After segmentation character recognition is done by introducing the method namely hybrid feature extraction with Multiclass SVM classification approach. The overall assessment of the research work is done in the matlab simulation environment and then it is proved the proposed HFE-MCSVM shows better performance.