PATTERN TRACKING ALGORITHMS USING SUCCESSIVE ABANDONMENT
Akikazu Kaga, Yoshio INOUE, Katsuhito Yamaguchi · Journal of Flow Visualization and Image Processing · 1993
A new algorithm for the pattern tracking velocimetry, which is much faster than the spatial correlation method, is proposed and applied to the measurement of low velocity airflow. In this algorithm the similarity of a pair of image fragments is estimated with the summation of gray level differences between corresponding pixels. The summation is executed in parts, and unpromising candidates are abandoned successively through F-inspection. The ratio of the number of vectors which are obtained as the unique solution to the total number of vectors, is defined as the reliability. This index increases as the fragment size increases, but the increasing rate is reduced as the quality of visualized images deteriorates. The algorithm discussed here is superior to other algorithms that have been proposed with respect to computation time and the ability to estimate reliability of results.