Title: Research on visualisation algorithm of handwritten digital image recognition based on deep neural network

Authors: Fang Teng; Xingliu Hu

Addresses: School of Art and Design, Shanghai University of Engineering Science, Songjiang District, Shanghai, China ' College of Intelligent Science and Control Engineering, Jinling Institute of Technology, Nanjing, Jiangsu, China

Abstract: The efficiency and accuracy of manual observations of Modified National Institute of Standards and Technology (MNIST) handwritten digital images are low. To solve this problem, a method based on a Deep Neural Network (DNN) model is proposed for screening, classifying and recognising handwritten digital images. MNIST handwritten digital images are used to train and test the DNN model for the rapid and accurate recognition. The average recognition accuracy of DNN model is 96.46%. The interactive interface is designed to realise the visualisation of programs and algorithms, and algorithms can be analysed from different angles and levels. From comparing the recognition effect of the DNN with Local Binary Pattern (LBP) feature extraction using texture features and edge feature extraction using shape features, the experimental results show that the DNN not only has high-recognition accuracy, but also simplifies the complex process for manually extracting image features.

Keywords: visualisation; handwritten digital image; image recognition; deep neural network; local binary pattern feature extraction; edge feature extraction; numbers; image classification; machine learning; visual interface.

DOI: 10.1504/IJCAT.2023.132552

International Journal of Computer Applications in Technology, 2023 Vol.72 No.1, pp.69 - 76

Received: 05 Aug 2022
Received in revised form: 24 Sep 2022
Accepted: 16 Oct 2022
Published online: 28 Jul 2023 *

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