In thermal spray process; the characteristics of in-flight particles (velocity and temperature) play an important role regarding the microstructure of the deposit and thus the coating performances. The implementation of diagnostic devices is necessary to measure such characteristics. Many imaging systems and algorithms have been developed for identifying and tracking in-flight particles. However; these current image systems have significant limitations in term of accuracy for example. One key to solving the tracking problem is to get an algorithm that can effectively distinguish different particles in the same image frame at the same time. This study aims to develop an algorithm capable of identifying a large number of in-flight particles sprayed by thermal process. Results show that the noise and vignettes could be successfully treated; particles are clearly recognized in the background; leading to properly measuring the sizes and positions of the particle versus time. The proposed algorithm has a higher recognition rate and recognition range than other algorithms; which will provide a reasonable basis for subsequent calculation and processing.