1994 年 14 巻 53 号 p. 108-115
A method to measure time average velocity distribution of fluctuating airflow through the digital image processing of multi-couples of images visualized by oil mist or smoke particles is proposed. Some fast algorithms for pattern tracking to avoid the increase of computation time with the increase of data amount are also proposed, and their performances are compared. It is shown through the comparison that the algorithm, in which the similarity of gray level patterns is defined with the accumulation of gray level differences between corresponding pixels, the accumulations are executed in parts and the candidates which differ significantly from the most promising candidate are abandoned successively through F-inspection, is the most reasonable one. In this algorithm, the reliability of calculated result is also obtained; it can be used as a measure to estimate the quality of visualized images.