A cupping spots image enhancement algorithm based on LAA-CycleGAN

AbstractDuring 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.

Source: European Journal of Applied Physiology - April 6, 2024 Category: Physiology Source Type: research