Performance of a U²-net model for phase unwrapping

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 U2-net has shown potential for various phase-unwrapping applications. This study proposes a U2-net-based phase-unwrapping model to explore the performance differences between the U2-net and U-net. To this end, first, the U-net, U2-net, and U2-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 U2-net model outperformed the U-net model. In particular, the U2-net-lite model achieved the same performance as that of the U2-net model while reducing the model weight size to 6.8% of the original U2-net model, thereby realizing a lightweight model.