Affiliation:
1. Guilin University of Technology
2. Second Affiliated Hospital of Guilin Medical University
Abstract
Abstract
During the automatic cupping process, a LAA-CycleGAN based image enhancement algorithm is proposed to address the issues of reduced image clarity and loss of detail features caused by fog adhering to the surface of the can body. Firstly, the generator contains a self-attention module to capture global features of the images; secondly, the discriminator introduces an Auto-Encoder to generate more stable images; Finally, a perceptual loss term is added to optimize the network for better perception. Experiments were conducted on the collected cupping spots dataset, and the results showed that compared with DCP, DehazeNet, AOD-Net, and CycleGAN algorithms, SSIM values increased by 48.78%, 61.02%, 53.45%, and 85.42%, while PSNR values increased by 5.02%, 5.09%, 4.78%, and 4.27%. The algorithm in this article reconstructs the cupping spots image with higher clarity, which can effectively enhance the quality of the cupping spots image and preserve details.
Publisher
Research Square Platform LLC
Reference22 articles.
1. A generalized dual-domain generative framework with hierarchical consistency for medical image reconstruction and synthesis[J];Zhang J;Communications Engineering,2023
2. Immediate and delayed effects of cupping therapy on reducing neuromuscular fatigue[J];Hou X;Frontiers in bioengineering and biotechnology,2021
3. Analysis of the influencing factors and clinical significance of pot spot[J];Ge Quanxi Ma;China Acupuncture,2018
4. Back acupoint location method based on prior information and deep learning[J];Liu YB;International Journal for Numerical Methods in Biomedical Engineering,2023
5. GSCYOLO: a lightweight network for cup and piston head detection[J];Liu YB;Signal, Image and Video Processing,2023