Table 1 The current research methods and limitations.

SRCNN

Apply super-resolution technology to enhance the resolution of panoramic radiographic images

The SSIM value reaches up to 0.98, and the PSNR value reaches up to 40.2 dB.

As the scaling factor increases, the model performance significantly deteriorates, and it is difficult to meet the specific requirements of digital art. The texture enhancement is not well coordinated with the artistic style.

²

RBFT

Improve the quality of underwater panoramic images through spatio-wave transformation technology

The SSIM value reaches up to 0.88, and the PSNR value reaches up to 31.2 dB.

Scenes with high turbidity or strong motion blur are prone to generating edge artifacts, and the processing speed is relatively slow. At the same time, it is difficult to restore the color gradation and texture details of the artistic scene, and the restoration results lack artistic coherence.

³

TRNet

Solving the problem of tripod occlusion in panoramic images based on generative adversarial networks

The SSIM value reaches up to 0.91, and the PSNR value reaches up to 32.5 dB.

The low-light restoration areas have many defects and lack a method for verifying the consistency of artistic style. The restored areas are prone to conflict with the overall style of digital art works, and their ability to generalize in artistic scenes is insufficient.

⁶

Two-step Approach

Utilize the end-to-end deep learning framework to solve the reflection problem of panoramic images

The SSIM value reaches up to 0.97, and the PSNR value reaches up to 38.5 dB.

The computational complexity is high, making it difficult to meet the real-time creation requirements of digital art. During the style conversion process, the artistic details are often distorted and fail to match the unique style characteristics of digital art.

⁷

YOLO-v5x

Repairing panoramic images using the YOLO network

Targetedly addressing the issue of dental panoramic image restoration

The restoration of the overlapping structure positioning has deviations, making it difficult to meet the requirements of artistic texture restoration. The generated results have low compatibility with the digital art texture style, and the defects can easily affect the artistic expression of the work.

⁸