A Novel Generalized Adaptive Domain Fusion Framework for Structured Light 3-D Measurement

Kepeng Qiu

DOI: https://doi.org/10.1109/jsen.2023.3340258

IF: 4.3

2024-01-01

IEEE Sensors Journal

Abstract:As a 3-D measurement technology that combines simplicity, accuracy, and noncontact measurement, structured light has attracted much attention in industrial metrology. Fringe projection profilometry (FPP) is one of the most promising structured light techniques based on phase analysis. In FPP, the phase-height model is a key element that significantly impacts the precision of 3-D measurements. However, the performance of such models is sensitive to the unwrapped phase and lacks the ability for adaptive adjustment. Despite extensive research on improving the mapping relationship of phase-height models, effectively using pixel phase information to establish a more accurate and reliable phase-height model remains a challenging issue, especially in cases with limited numbers of calibration planes. To address this issue, we propose a generalized adaptive domain fusion (GADF) framework for FPP. The core principle of the GADF framework is to use a specially designed dual-indexing criterion (DIC) to mine the adaptive domain for each pixel and then use the domain model fusion technology to improve the accuracy of the height measurement. Experimental results demonstrate that the GADF framework outperforms traditional phase-height models.