Abstract: To address the lack of morphological detection technologies for smoke aerosols, this study developed a biomimetic aerosol dynamic characterization system based on the human oral cavity structure. By integrating synchronized laser imaging and multi-parameter quantitative analysis, the system enables spatiotemporal visualization and morphological feature extraction of aerosol diffusion. Aerosol behavior under varying initial velocities （modulated by adjusting cigarette diameter） was investigated through the extracted four key parameters from morphological images: projection area （S）, global mean grayscale （L）, regional grayscale standard deviation （σ_L）, and centroid migration （C_x, C_y）. The results show that the relative deviation for ten repeated detections of S is 0.9%, demonstrating the stability of the device. The positive correlations between inlet initial velocity and S （ρ=0.96）, L （ρ>0.9）, and σ_L （ρ=0.94） are also revealed. The dynamic evolution of these feassures aligned with an attenuation-deposition hybrid model, demonstrating high-velocity aerosols in curved-wall confined spaces exhibit the reduced sedimentation, improved uniformity, and intensified clustering. The developed platform provides a robust and precise toolkit for the research on the aerosol diffusion dynamics and deposition characteristic.