The interactive embodiment of aesthetics combined with visual object recognition algorithm in graphic design in the field of artificial intelligence

Visual aesthetic impact has become the primary embodiment of interactivity in graphic design with the rapid development of internet technology and the deep promotion of modern graphic design concepts. Based on the combination of visual object recognition algorithms and interactive features in graphi...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:International journal on interactive design and manufacturing Ročník 18; číslo 4; s. 2517 - 2528
Hlavní autor: Wang, Jing
Médium: Journal Article
Jazyk:angličtina
Vydáno: Paris Springer Paris 01.05.2024
Springer Nature B.V
Témata:
Accuracy
Aesthetics
Algorithms
Artificial intelligence
CAE) and Design
Clustering
Color vision
Computer graphics
Computer-Aided Engineering (CAD
Crowdsourcing
Design
Designers
Electronics and Microelectronics
Engineering
Engineering Design
Experiments
Graphic design
Industrial Design
Innovations
Instrumentation
Labeling
Labels
Mechanical Engineering
Object recognition
Original Article
Prediction models
Semantics
Visual fields
Deep learning
Graphic design
Interactive research
Visual object recognition
Artificial intelligence
Feature modeling
ISSN:1955-2513, 1955-2505
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Visual aesthetic impact has become the primary embodiment of interactivity in graphic design with the rapid development of internet technology and the deep promotion of modern graphic design concepts. Based on the combination of visual object recognition algorithms and interactive features in graphic design in the field of artificial intelligence, this research aims to meet users’ needs for diverse graphic design images through intelligent generation. First, a hierarchical model is proposed to objectively quantify the features of graphic design images. The model divides images into pixels, elements, relations, planes, and application levels. It defines the representation of features for each category. Element colors are extracted using a clustering algorithm to calculate layout and color perception features. The model maintains proper academic structure and language conventions while providing clear and concise information. Next, a prediction model is constructed for image geometric features in graphic design. By fitting the distribution of geometric features found in the data, a prediction is made that the probability distribution of position and color elements under uncertain conditions will facilitate the production of flat images. To create the predictive model, the method of sampling features was investigated. Finally, the graphic design is reconstructed according to the interactive features to generate diversified graphic design images. In 3000 random graphic design images, each feature is grouped and visualized as clustered. The results show that the images with more clusters are influenced by the background color, and the images with the same color system are easier to cluster. The test results imply that the accuracy rates of design elements, design labels and label borders are 99.7%, 89.5% and 83.2% respectively, indicating that in the existing labeling results, there are very few labeling errors of design elements case. By combining visual interactive displays to construct graphic design images with different feature distributions through feature quantification, graphic design images with different styles and features be effectively generated to meet the individual needs of different consumers.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1955-2513
1955-2505
DOI:10.1007/s12008-023-01723-9