Journal of South China University of Technology (Natural Science Edition) ›› 2020, Vol. 48 ›› Issue (3): 1-9.doi: 10.12141/j.issn.1000-565X.190442

• Electronics, Communication & Automation Technology •     Next Articles

User Image Aesthetic Representation Based on Personalized Generation Template Aggregation

WANG Weining1 MA Xuedong1 SU Junjie1 LUO Jiebo2 XU Xiangmin1   

  1. 1. School of Electronic and Information Engineering, South China University of Technology, Guangzhou 510640, Guangdong,China; 2. Department of Computer Science, University of Rochester, Rochester 14627, New York, USA
  • Received:2019-07-11 Revised:2019-10-23 Online:2020-03-25 Published:2020-03-01
  • Contact: 徐向民（1972-），男，教授，主要从事计算脑科学、人工智能、人机交互、柔性穿戴与智能集成系统等研究。 E-mail:xmxu@scut.edu.cn
  • About author:王伟凝（1975-），女，副教授，主要从事计算机视觉、图像情感分析、图像分类与检索、机器学习研究。E-mail:wnamg@scut.edu.cn
  • Supported by:
    Supported by the National Natural Science Foundation of China (U1801262, U1636218, 61702192), the State Scholarship Found of China ( 201506155081) and the Natural Science Foundation of Guangdong Province (2015A030313212)

PDF

138

Abstract: The general image aesthetic research can not represent the user’s personalized aesthetic preferences becaused it does not consider the aesthetic differences among users. In order to represent the user’s personalized aesthetic preferences more effectively, a user aesthetic representation construction method based on personalized generation template aggregation using the Counting Grid model was proposed in this research. It constructs a more compact user image aesthetic representation vector, which can better describe the user characteristics. Experimental results show that the way of user aesthetic representation constructed by the proposed method can obtain higher recognition rate in user identification, and can also obtain more realistic recommendation results in user recommendation.

Key words: aesthetic quality, image feature, personal research, user identification, recommended system

CLC Number: