Abstract
Conventional PIV method in which cross correlation coefficient is used has problem that erroneous vectors occur considerably if the correlation matrix is reduced to raise the spatial resolution. In present algorithm, searching for image in correlation matrixes and dividing the matrix are iterated, and after that the second step image transforms to approximate to the first step by using obtained vectors. Shift vectors at each point are obtained by repeating these operation until satisfying necessary space resolution. It is possible to search directly for a position where a correlation coefficient indicates the maximum value, because the shift distance is always smaller than the correlation matrix size. This method is applied to the standard PIV images and the visualized images of wake behind a circular cylinder. And from the results of these PIV measurement, it made clear that initial number of image division must choose appropriate number to prevent emerging erroneous vectors.
