An Adaptive Denoising Algorithm for Speckle Noise in DoFP Polarization Images
Abubakar Abubakar, Amine Bermak · 2021
In this paper, an adaptive algorithm for filtering speckle noise in Division-of-Focal-Plane (DoFP) polarization images is presented. The proposed algorithm involves replacing the first round of operation in the Block-matching and 3D Filtering (BM3D) denoising algorithm with an adaptive thresholding technique based on the smoothness of an image patch. This improves the "Semi-Filtered" image that is passed to the final BM3D denoising round. With this adaptive technique, key image details are preserved and speckle noise is significantly suppressed. Experimental results on four DoFP test images indicate the significant performance of the proposed algorithm in suppressing speckle noise. This significant performance is then compared with the performance of other existing denoising algorithms. Comparison is done numerically using Peak-Signal-to-Noise-Ratio (PSNR) and Structural Similarity (SSIM) Index scores. Thereafter, comparison is extended to visual results to ascertain agreement between the two sets of comparisons.