Applied Mathematics and Nonlinear Sciences
Journal license

Journal

Applied Mathematics and Nonlinear Sciences


Volume
& Issue

Volume 9, Issue 1


Published
on

November 27, 2024


Pages


DOI

Article

Research on Intelligent Extraction and Generative Design Creation of Cultural Heritage Image Elements Based on Visual Recognition and Machine Learning

Check for updates


Authors

Lingyun Chen Affiliation:
College of Art and Design, Shanghai Polytechnic University, Shanghai, 201209, China


Abstract

In this paper, machine learning and computer vision image segmentation methods are used to achieve classification of images and recognition and extraction of pattern elements. It also extracts colors from the imagery using intelligent image processing techniques, such as preprocessing and color clustering of Tangshan shadow images, and displays the results of color resolution. The accuracy of this paper’s method is reflected in the recognition and extraction of cultural heritage food images, as shown in the results. Tangshan shadow image elements are clustered into five categories, the proportion of colours used in Xiaodan are all between 0~12.5%~25%, and the proportion of high saturation in all five categories is more than 60%. The highest percentage of high brightness in Xiaodan colors is 36.2%. Applying Tangshan shadow image elements and color features to the brand logo and packaging design, a significant portion of respondents said they were very satisfied with the visual effect. The image elements and applications of cultural heritage also provide new perspectives and theoretical basis for modern visual design.


Keywords

Machine learning, Elements, Colour features, Image segmentation, Recognition and extraction., 97P10


Citation

Chen, L. (2024). Research on intelligent extraction and generative design creation of cultural heritage image elements based on visual recognition and machine learning. Applied Mathematics and Nonlinear Sciences, 9(1). https://doi.org/10.2478/amns-2024-3551
0 Total citations
0.00 FWCI
0 Recent citations
(2 years)
20 References
Open Access Yes
View full metrics

Published by: Engineering Journals

Engineering Journals Logo