Panoramic image restoration and visual quality enhancement methods for digital art creation

Area of Science:

• Computer Vision
• Digital Image Processing
• Artificial Intelligence

Background:

• Panoramic images are vital for digital art and virtual reality but current restoration methods lack semantic understanding and global control.
• Existing techniques often result in structural disorder and struggle to balance comprehension, performance, and quality.

Purpose of the Study:

• To develop a panoramic image restoration and visual quality enhancement model for digital art creation.
• To address limitations in semantic understanding, global feature control, and performance efficiency in current methods.

Main Methods:

• A Multi-Scale Residual Network, Coordinate Space Attention, and super-resolution reconstruction form the core enhancement algorithm.
• An optimized Generative Adversarial Network and Vision Transformer model spatial correlation and semantic logic for high-quality completion.

Main Results:

• The model achieved high performance metrics, including a Structural Similarity Index (SSIM) of 0.975 and Peak Signal-to-Noise Ratio (PSNR) of 53.82 dB.
• It demonstrated superior performance over comparison models in detail clarity, global consistency, and processing efficiency (3.12s response time).

Conclusions:

• The proposed model effectively enhances panoramic image quality for digital art creation.
• It offers significant advantages in visual quality, detail preservation, global consistency, and processing efficiency.