Accurate fringe order recognition with adaptive morphological operations for 3D measurement

Abstract

Temporal phase unwrapping (TPU) techniques are vulnerable to inaccurate recognition of fringe order due to various factors, including noise, motion blur and defocusing effects. The misalignment between the fringe order and wrapped phase can lead to abnormal jump errors and compromise measuring accuracy. To address this issue, an accurate fringe order recognition method with adaptive morphological operations for 3D measurement is proposed. Differs from the traditional half-period fringe order recognition method, a complete full-flow algorithm is constructed to accurately separate adhesive connected domains, which includes: segmentation and erosion of connected domains, localization of fully eroded connected regions, clustering of separated valid points, and adaptive selection of radius. In contrast to a simplistic morphological open arithmetic, the proposed method has the capability to perform clustering on all separated valid points, thereby achieving lossless fringe order recognition of the full field. And its row-correlation property gives it a larger correctable range than the previous row-independent approach. The experimental results demonstrate the generality, advancement, adaptivity, and precision of our method, built upon the traditional TPUs for obtaining fringe orders.