BM3D Filtering Algorithm for DoFP Polarization Image Sensors

In this paper, we present a Block Matching and 3D filtering (BM3D) algorithm dedicated to the division-of-focal plane (DoFP) polarization image sensors. This method, based on grouping and a non-local filtering method named collaborative filtering, exploits all the different polarization channels simultaneously. Extensive experimental results on the test images show that the algorithm outperforms a wide range of existing denoising algorithms for DoFP images including the state-of-the-art principal component analysis (PCA) based denoising algorithm visually by preserving more details, as well as in terms of Peak-Signal-to-Noise-Ratio (PSNR).