Proceedings of the 13th EAI International Conference on Mobile Multimedia Communications, Mobimedia 2020, 27-28 August 2020, Cyberspace

Research Article

Research on image sensitive information recognition based on machine learning

Download641 downloads
  • @INPROCEEDINGS{10.4108/eai.27-8-2020.2294703,
        author={Kaiyong  Li and Duo-lu  Mao},
        title={Research on image sensitive information recognition based on machine learning},
        proceedings={Proceedings of the 13th EAI International Conference on Mobile Multimedia Communications, Mobimedia 2020, 27-28 August 2020, Cyberspace},
        publisher={EAI},
        proceedings_a={MOBIMEDIA},
        year={2020},
        month={11},
        keywords={machine learning; image sensitive information; identification method; support vector machine;},
        doi={10.4108/eai.27-8-2020.2294703}
    }
    
  • Kaiyong Li
    Duo-lu Mao
    Year: 2020
    Research on image sensitive information recognition based on machine learning
    MOBIMEDIA
    EAI
    DOI: 10.4108/eai.27-8-2020.2294703
Kaiyong Li1,*, Duo-lu Mao1
  • 1: Qinghai Nationalities University Physics and Electronic Information Engineering, Xining, 810007 China
*Contact email: weiyi220@tom.com

Abstract

Aiming at the problem of low accuracy in traditional image sensitive information recognition methods, a new image sensitive information recognition method based on machine learning was proposed. The pre-processing operations of de-noising and detail sharpening are carried out for the recognition image, and the pre-processing image features are extracted from the three perspectives of color, shape and texture. The image sensitive information is retrieved by combining the extracted image features. Based on the principle of support vector machine (SVM) in machine learning, an image classifier is designed to realize the classification and recognition of image sensitive information. By comparing with two traditional image recognition methods, it is proved that the recognition method based on machine learning proposed in this paper has higher precision and is more suitable for identifying sensitive information of images.