Title: Fast recognition method for handwritten Chinese character text based on multi granularity convolution

Authors: Pingge Huang; Xinhua Wang

Addresses: Department of Basic Teaching and Research Shangqiu Polytechnic, Shangqiu, 476000, China ' Department of Criminal Science and Technology, Henan Police College, Zhengzhou, 450046, China

Abstract: Based on the research motivation of solving the problems of low accuracy, high misidentification rate, and long time in current handwritten Chinese character text recognition methods, a new fast recognition method for handwritten Chinese character text based on multi granularity convolution is proposed. Enhance handwritten Chinese character text data to expand the size of the training dataset. Based on the enhanced data, the handwritten Chinese character text data is subjected to binarisation, median filtering, and segmentation processing. The segmented handwritten Chinese character text is input into a multi granularity convolutional neural network, which iteratively calculates and outputs fast recognition results for handwritten Chinese character text. The experimental results show that the maximum accuracy of the proposed method for handwritten Chinese character text recognition is 99.13%, the minimum false recognition rate is 1.44%, and the minimum recognition time for handwritten Chinese character text is 0.23s, which meets the research expectations.

Keywords: multi granularity convolution; handwritten Chinese character text; fast recognition; binarisation; median filtering; segmentation processing.

DOI: 10.1504/IJBM.2025.149697

International Journal of Biometrics, 2025 Vol.17 No.6, pp.596 - 614

Received: 11 Mar 2025
Accepted: 07 Aug 2025
Published online: 10 Nov 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article