Denoising of Smart-phone based Fringe Projection Image using Curvelet Transform

Fourier transform profilometry (FTP) can enable three-dimensional(3-D) surface profiles through single-shot imaging, which make it one of the popular high-throughput, non-contact topography methods. Here, we developed a portable FTP using a smart phone and a micro projector. The micro projector casts a fringe pattern onto the measured object surface and the smart phone captures the resulting deformed fringe image. However, the recorded image has distortion and noise, which seriously degrade the reconstructed quality. In addition, the traditional transform, such as Fourier and wavelet, can suppress parts of high frequencies yielded by sharp changes in shape, resulting in image blurring. We then introduce Curvelet transform to inhibit the noise and preserve the high frequencies owing to its multiscale and multi-direction analysis in image processing capacities. Simulations and the experimental results show that the Curvelet transform has an excellent performance in image denoising and Curvelet transform denoising achieves high-precision 3D reconstruction.