Abstract: To deeply understand dynamic evolutionalmechanism of initial of aeolian sand particles by multiple factors, it isvery critical to measure precisely liftoff velocities and trajectories ofsand particles with different movement types. Although the high-speedphotography illuminated by intensive and continuous laser is an effectivemethod for obtaining the above information, the kinds of high- speedphotography images cannot be processed precisely by overlapping image methoddue to very low contrast ratio of objectives to backgrounds and significantdiscrepancy between two frame images. As a result, a new arithmetic methodof image segmentation is proposed on basis of a principle of variation ingray level difference between two frame consecutive images. The cases showthat the new arithmetic method can segment successfully saltating particlesfrom immobile bed by selecting appropriate threshold value of gray leveldifference between two continuous images and by adopting adaptive binarymethod. The new arithmetic method plus a minimum distance matching method ofparticles tracking velocimetry (PTV) can retrieve precisely sand particlestrajectories near the bed as sand particles concentration is low.