Moment of Inertia-Based Approach to Recognize Arabic Handwritten Numerals

Autor:	Binod Kumar Prasad
Rok vydání:	2019
Předmět:	Numeral system Pixel business.industry Handwriting recognition Computer science Feature extraction Pattern recognition Artificial intelligence Moment of inertia business Hidden Markov model MNIST database Coding (social sciences)
Zdroj:	Lecture Notes in Networks and Systems ISBN: 9789811337642
DOI:	10.1007/978-981-13-3765-9_26
Popis:	Simpler methods of feature extraction and better accuracy have always been primary needs of a handwriting recognition system to be a successful real-time system. The novelty of this paper lies in the introduction of two unique methods of feature extraction which are Pixel Moment of Inertia (PMI) and Delta Distance Coding (DDC). PMI absorbs angular variations in the samples and DDC performs precised local curve coding for better recognition accuracy. Multiple Hidden Markov Model (MHMM) has been used to neutralize the effect of two very frequent writing styles of numerals “4” and “7” on their recognition rates. The paper uses MNIST database, and overall recognition accuracy of 99.01% has been achieved.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::7ec24eb29ff3fc47e7543d185247228f https://doi.org/10.1007/978-981-13-3765-9_26 Zobrazit plný text záznamu