Journal of Beijing University of Posts and Telecommunications

  • EI核心期刊

Journal of Beijing University of Posts and Telecommunications ›› 2021, Vol. 44 ›› Issue (6): 53-58.doi: 10.13190/j.jbupt.2021-097

• PAPERS • Previous Articles     Next Articles

Evaluation Method of Traditional Pattern Vector Diagram Based on Perception Drive

ZHAO Hai-ying, LI Huan-hong, HOU Xiao-gang   

  1. School of Artificial Intelligence, Beijing University of Posts and Telecommunications, Beijing 100876, China
  • Received:2021-06-07 Online:2021-12-28 Published:2021-12-28

Abstract: Aiming at the question of how to evaluate the quality of vector materials in the construction of the Chinese cultural vector material library, three visual evaluation indicators of accuracy, continuity and simplicity are proposed and quantitatively analyzed. First, the vector diagram is analyzed to obtain the document object model of the vector diagram, and the document tree is traversed to obtain the vector information. Then, by combining the Gestalt psychology, the visual error, curve continuity, and complexity of the decorative pattern of the vector diagramis quantified. Finally, an evaluation method based on perception-driven traditional pattern vector diagrams is proposed. The proposed algorithm provides three quantitative indicators that can be used for automatic evaluation, sorting and screening of vector diagrams in a multitask-driven vectorization system. Simulation results on the texture vector data set show that the evaluation results of the vector graphics meet the visual aesthetics and expectations of mainstream designs.

Key words: traditional patterns, visual perception, image vectorization, evaluation method

CLC Number: