Synthesize-Then-Evaluate Based Optimal Exposures Selection for High Dynamic Range Fringe Projection Profilometry

Fringe projection profilometry (FPP) has been widely applied in many industrial scenes, e.g., quality inspection and maintenance of automobile manufacturing. However, the practical application of FPP faces challenges on high dynamic range (HDR) surfaces of industrial parts. Existing HDR approaches tends to preset exposure sequences on mere analysis on pre-captured images, which is inflexible and inefficient. This article proposes an optimal exposure selection approach based on synthesize-then-evaluate. Unlike existing approaches, in our approach, the optimal exposure sequence is formulated as an iterative optimization problem and an iterative exposure selection framework is proposed to tackle it. A phase evaluation metric based on phase-encoded information quantity is devised to thoroughly assess the image exposure quality. Besides, a synthesize-then-evaluate model is established to efficiently and iteratively search for the optimal exposure in each iteration optimization problem. To further enhance the efficiency of the process, this model integrates Bayesian optimization and exposure image synthesis method based on the camera response function. Experimental results with various industrial parts indicate that, the proposed approach can significantly reduce the number of required exposures while maintaining equivalent surface coverage rates.