High-dynamic-range 3D measurement with temporal line-shifting structured light and saturated-frame-count-guided multi-exposure fusion

Non-Lambertian high-dynamic-range surfaces often exhibit local under- and overexposure, causing fringe loss and reconstruction errors. We propose a saturated-frame-count-guided multi-exposure fusion method for temporal line-shifting structured light. High and low exposures are selected using percentile gray levels and image entropy, and only pixels exhibiting high-exposure errors are replaced. Step-block tests yielded maximum mean absolute error (MAE) and root-mean-square error (RMSE) of 0.0628 and 0.0742 mm, respectively. Compared with sinusoidal multi-exposure fusion, high-exposure retention increased from 2.9719% to 42.7275% in bright regions and from 78.2437% to 97.2410% in dark regions, reducing redundant replacement without pixel mapping or intensity modulation.