Accurate particle three-dimensional diagnostic in noisy combustion environments through holographic similarity analysis

In situ particle characterization is crucial for experimentally evaluating multiphase flow systems, where digital holography offers unique three-dimensional diagnostics. However, reliable particle localization remains challenging under strong optical noise encountered in practical environments. This study introduces a similarity-driven autofocusing framework that reformulates depth sensing by leveraging inter-particle holographic similarity, quantified by mutual information (MI), as a robust, collective focus criterion. Through systematic benchmarking against seven conventional algorithms, MI demonstrates superior localization accuracy under diverse noise conditions (Gaussian, Speckle, and Salt and Pepper noise at signal-to-noise ratio ∼ 10 dB). The core innovation lies in the first application of MI to establish a noise-immune focus signature from correlated particle-pair propagation. Experimental validation on noisy holograms of combusting metal particles confirms the method's efficacy. This work bridges information theory and computational optics, establishing a new approach for reliable particle tracking in combustion diagnostics, aerospace propulsion, and industrial spray characterization.