Hybrid algorithm for image denoising of mixed speckle-impulse noise in planar laser-induced fluorescence (PLIF) imaging

Abstract

Flow visualization is essential for a comprehensive understanding of complex fluid dynamics. Planar laser-induced fluorescence (PLIF) has emerged as one of the most promising techniques for multi-dimensional imaging acquisition, offering high spatial and temporal resolutions. However, PLIF images frequently suffer from impulse and speckle noise under low-density, low-pressure conditions, thereby affecting the quality of visualization. This study proposes a novel hybrid algorithm that integrates order-statistic and non-local means filters with residue feedback. The algorithm enhances image quality and eliminates noise effectively by incorporating recursive processes guided by the structural similarity index (SSIM) and anisotropic diffusion. The efficacy of the method is confirmed through evaluation using peak signal-to-noise ratio (PSNR) and SSIM on images with varying noise levels. The algorithm demonstrates computational efficiency and shows promise for denoising PLIF images.