Fringe pattern despeckling based on homomorphic transform and enhanced low-rank matrix approximation

To address the accuracy degradation in fringe projection 3D measurement caused by strong laser speckle noise, this paper proposes a speckle suppression method based on homomorphic transform combined with enhanced low-rank matrix approximation. Firstly, the multiplicative speckle noise in structured-light images is converted to approximately Gaussian additive noise via homomorphic transformation. Noise suppression is then performed in the transformed domain using enhanced low-rank matrix approximation. Finally, the denoised fringe pattern is obtained by applying the inverse homomorphic transformation. We also demonstrate the practicality of our proposed method through simulations and experimental results.