Abstract
Phase unwrapping plays a pivotal role in optics and is a key step in obtaining phase information. Recently, owing to the rapid development of artificial intelligence, a series of deep-learning-based phase-unwrapping methods has garnered considerable attention. Among these, a representative deep-learning model called ${{\rm U}^2}$-net has shown potential for various phase-unwrapping applications. This study proposes a ${{\rm U}^2}$-net-based phase-unwrapping model to explore the performance differences between the ${{\rm U}^2}$-net and U-net. To this end, first, the U-net, ${{\rm U}^2}$-net, and ${{\rm U}^2}$-net-lite models are trained simultaneously, then their prediction accuracy, noise resistance, generalization capability, and model weight size are compared. The results show that the ${{\rm U}^2}$-net model outperformed the U-net model. In particular, the ${{\rm U}^2}$-net-lite model achieved the same performance as that of the ${{\rm U}^2}$-net model while reducing the model weight size to 6.8% of the original ${{\rm U}^2}$-net model, thereby realizing a lightweight model.
© 2023 Optica Publishing Group
Full Article | PDF ArticleMore Like This
Jie Chen, Yong Kong, Dawei Zhang, Yinghua Fu, and Songlin Zhuang
Opt. Express 31(18) 29792-29812 (2023)
Jiaxi Zhao, Lin Liu, Tianhe Wang, Xiangzhou Wang, Xiaohui Du, Ruqian Hao, Juanxiu Liu, Yong Liu, and Jing Zhang
Opt. Express 30(22) 39794-39815 (2022)
Hongxuan He, Chen Tang, Liao Zhang, Min Xu, and Zhenkun Lei
J. Opt. Soc. Am. A 40(10) 1969-1978 (2023)