Title: Edge denoising of art illustration image based on contour feature recognition

Authors: Wei Zhao

Addresses: Department of Fine Arts, Mudanjiang Normal University, Mudanjiang 157011, China

Abstract: In order to solve the problems of low precision and long time-consuming in traditional edge denoising methods of art illustration image, an edge denoising method of art illustration image based on contour feature recognition is proposed. The edge of art illustration image is segmented, and the feature target value and background value are extracted. By calculating the same degree of edge data in each kind of features, the edge feature classification of art illustration image is realised with the help of naive Bayes classification matrix. The edge noise region of the image is determined, and the wavelet descriptor in the contour descriptor is used to smooth the edge noise region of the art illustration image to complete the edge denoising of the art illustration image. Experimental results show that the edge denoising accuracy of the proposed method is about 95%, and the denoising time is only 2.1 s.

Keywords: contour features; art illustration images; edge denoising; feature extraction.

DOI: 10.1504/IJART.2021.10043884

International Journal of Arts and Technology, 2021 Vol.13 No.4, pp.355 - 366

Received: 18 Dec 2020
Accepted: 10 Oct 2021

Published online: 23 Feb 2022 *

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