Open AccessHandwritten digit recognition based on corner detection and convolutional neural network
Author(s) -
Chen Zhong,
Yuhang Wang,
Decheng Zhang,
Wang Ke
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1651/1/012165
Subject(s) - computer science , convolutional neural network , artificial intelligence , pattern recognition (psychology) , numerical digit , set (abstract data type) , digit recognition , data set , test set , speech recognition , artificial neural network , field (mathematics) , image (mathematics) , test data , handwriting recognition , feature extraction , mathematics , arithmetic , pure mathematics , programming language
In the field of inspection and testing, it is necessary to manually fill a large number of test record forms, and the digital information accounts for a large proportion in the record forms. For the purpose of automatic and accurate recognition of handwritten form data, this paper proposes a handwritten digit recognition method based on Harris corner detection algorithm and convolutional neural network(CNN). The Harris corner detection algorithm is used to identify the positioning marks in the form image, which determine the possible fill-in positions of handwritten digits. Through a 14-layer CNN model, and using the ReLu function as the excitation function, the handwritten digit features of the target area in the image are recognized. The experimental results show that the average recognition accuracy rate of this method on the test set can reach 98.14%. So a technical solution for intelligent recognition of record forms is thus provided.