Research on Digital Media Intelligent Art Creation Based on the Fusion of Virtual Reality and Semantic Features

YANG LIU ； CUN SHANG FAN ； PENG-XIN LIU

《Journal of Information Science and Engineering》 39卷1期 (2023/01) Pp. 55-66

https://doi.org/10.6688/JISE.202301_39(1).0004

摘要

With the rapid development of computer science, technologies such as virtual reality and semantic feature fusion have been greatly developed in recent years. In order to construct a digital media intelligent art creation model, this paper first designs a digital media intelligent creation framework based on virtual reality and maintains the original pattern and color distribution of natural images as accurately as possible. This method first performs automatic color segmentation and clustering on the image to extract key lines; then, re-colors the different segmented regions; and finally completes the detail enhancement. For the optimization of color matching, the algorithm integrates the contrast of the color block, and uses the differential evolution strategy to optimize the color configuration scheme. In addition, this paper also proposes a generative adversarial network-based on semantic feature fusion for digital media intelligent art creation. The network introduces multi-scale feature fusion by embedding the residual block feature pyramid structure and directly generates the final fine image by adaptively fusing these features. Only one discriminator can generate a realistic image of 256px × 256px. The experimental results have shown that this paper makes full use of virtual reality and semantic feature fusion technology, and obtains better results and better performance of intelligent art creation results on digital media than existing methods.

關鍵字

virtual reality ； semantic feature fusion ； digital media ； intelligent art creation ； computer vision