High-dynamic-range 3D shape measurement: the original and inverted fringe projection method combined with the K-means clustering exposure time estimation algorithm

It is a challenge for fringe projection profilometry to reconstruct the three-dimensional (3D) shape of highly reflective objects due to the saturation of the captured images. In the multiple exposure method (MEM), exposure time is not predetermined, which requires numerous adjustments to achieve a wide range of intensity fringe sequences. This results in low measurement efficiency. To address this, this paper proposes an original and inverted fringe projection method (OIFPM) combined with a K-means clustering exposure time estimation algorithm. First, we apply the K-means clustering algorithm to classify the intensity levels of each pixel in the captured image, which allows for the estimation of the optimal exposure time. Subsequently, we combine the above method with OIFPM to capture original and inverted fringe images at each estimated exposure time and select the maximum intensity but unsaturated pixels to synthesize the final fringe image. Finally, the phase calculation is performed using 72 wrapped phase equations, and these equations are presented in detail. Experimental results demonstrate that the proposed approach performs excellently in the 3D measurement of highly reflective objects. It significantly outperforms both OIFPM and the random triple exposure time method, achieving results comparable to those of MEM while requiring fewer fringe images. This makes the proposed method particularly suitable for the 3D measurement of highly reflective objects.