Relaxation method based PTV with bidirectional principle

The algorithm for RM-PTV and its inefficiency to calculate "isolated particle being no match" resulting in illogical match were introduced, and the bidirectional principle was accordingly proposed. Certain double-frame image pairs as self-defined flow field for examination were produced. Moreover, controlled particle deletion and addition were adopted in those second frames to simulate "particle being no match" and noise, respectively. Unidirectional and bidirectional RM-PTV were then used successively to calculate their particle matching. The results uniformly show that unidirectional RM-PTV performs well against noise, but is impacted severely by "isolated particle being no match", whereas the bidirectional RM-PTV could keep the accuracy η consistently high in a rather bad condition. The result of vortex flow field with spatial rate Pv being 14.563 is shown as example that change of noise proportion μa hardly influence η in unidirectional method, but its η decreases to about 80% when "isolated being no match proportion" μd increases to 30%, comparatively the bidirectional method keeps η at 95% even when μa and μd are both 30%. Then the result of standard synthetic image pair using both methods validated again the ability of bidirectional principle in eliminating illogical match.