Dynamic 3-D Measurement Based on Temporal Fringe Order Recognition

Efficient and robust fringe order recognition is an important topic in 3-D measurement. In this article, a novel dynamic 3-D measurement method based on temporal fringe order recognition is proposed. Using binary defocusing technology, the high-speed measurement system can cyclically project and capture three-step phase-shifting fringe patterns at 120 frames/s, enabling computation of the wrapped phase from three adjacent deformed patterns. Due to the ultra-short interval, the phase relationship between each interval is explored and established by segmenting the connected domain. With coarse and fine recognition, the temporal fringe order at each interval can be accurately recognized along the time axis. The proposed method can achieve a dynamic 3-D reconstruction rate of 120 frames/s while also possessing a similar single-shot feature. Moreover, our temporal fringe order recognition method can realize phase unwrapping without any additional pattern, except for the projection pattern in the first frame reconstruction result. It provides a unique solution to the challenges posed by traditional temporal phase unwrapping (TPU) methods in which additional patterns must be required. High-quality 3-D results of dynamic scenes, including isolated objects in motion, left-right motion with a mask held in hand, rotational motion of a Davidian object, and hand-grasping paper, were presented at a frame rate of 120 frames/s. Experimental results demonstrate its effectiveness and low cumulative effect of spatial and temporal errors.