Wavelet analysis for shadow detection in Fringe Projection Profilometry

Improving 3D reconstruction accuracy in wavelet transform profilometry by reducing shadow effects

“…Fringe projection technique (FPT), specifically, is a structured light pattern technique, which detects the surface elevation of the object to be reconstructed through the distortion of a fringe pattern [3, 4]. However, during the capture process, shadows are generated by object occlusion.…”

“…The phase‐unwrapping algorithm used in [13], Herráez, was one of five different algorithms evaluated in the present paper [14]; the results using the Bioucas [15] and Ghiglia [16] algorithms were superior. In [4], the Haar wavelet analysis used the sharp variations in intensity at the shadow boundaries, together with the mean and standard deviation of fringe intensity between fringe peaks, to detect shadow boundaries, even under conditions with random noise present. However, Hani et al [4] propose that phase unwrapping and reconstruction do not take place in shadow areas, rather than the novel approach of replacing shadow areas with background reference pixels, as proposed in the present paper.…”

“…In [4], the Haar wavelet analysis used the sharp variations in intensity at the shadow boundaries, together with the mean and standard deviation of fringe intensity between fringe peaks, to detect shadow boundaries, even under conditions with random noise present. However, Hani et al [4] propose that phase unwrapping and reconstruction do not take place in shadow areas, rather than the novel approach of replacing shadow areas with background reference pixels, as proposed in the present paper. The simple, rectangular object used in [4] was not evaluated for the final reconstruction quality after phase unwrapping, while the present paper uses six non‐trivial objects and presents numerical and visual results for each.…”

“…However, Hani et al [4] propose that phase unwrapping and reconstruction do not take place in shadow areas, rather than the novel approach of replacing shadow areas with background reference pixels, as proposed in the present paper. The simple, rectangular object used in [4] was not evaluated for the final reconstruction quality after phase unwrapping, while the present paper uses six non‐trivial objects and presents numerical and visual results for each. In [17], two methods are proposed for shadow detection using photometric stereo; however, these involve the use of two cameras and employ pixel‐intensity comparisons.…”

Investigation of accurate fringe analysis techniques for fringe projection profilometry

“…Since no signal is detected in the shadow or glare region, Wavelet spectrum will be zero, thus allowing us to determine the shadow areas accurately. [157].…”