Five techniques for increasing the speed and accuracy of PIV interrogation

This: paper explores several methods for increasing the efficiency and accuracy of particle image velocimetry (PTV) data analysis. The time to directly compute the correlation function to determine the displacement in PIV interrogation windows is reduced using two techniques. First, a scheme that calculates the correlation for 16 pixels in parallel is implemented. A further increase in efficiency results from using a truncated multiplication scheme that calculates 82% of the answer while reducing the work by 84%. Second, advantage is taken of the common practice of overlapping adjacent interrogation windows by not recorrelating the portion of the new window that overlaps the old. For the commonly used 50% overlap, the speed of vector calculation can theoretically be increased 300%, In practice, the improvement depends on the implementation of the method. The most efficient algorithm doubles the processing speed at 50% overlap. Accuracy is increased using three methods. First, a technique for enhancing PIV images to increase the contrast between the particles and the background is presented. This method is particularly useful when experimental exigencies result in low contrast images. Second, a method is presented for resolving the velocity in areas of high velocity gradient where the correlation map contains multiple peaks. Third, equalizing the histogram of sub-pixel adjustments should eliminate peak locking. Sample data show a 'decrease' in error of 15%.