In a digitally recorded video image, an area of interest of the flow field is selected and divided into subsample areas known as areas of interrogation (Figure 1). Each area of interrogation is compared at Time 1 (T1) with the corresponding area at Time 2 (T2) in the following video frame. In our studies of fish wakes, the time elapsed between frames is typically 4 ms. In order to determine the path taken by the particles from T1 to T2, a statistical cross-correlation is performed. A high cross-correlation value is given when the particle images at T1, when shifted in space by jogging the entire area of interrogation, match most of the particle images at time 2. The peak cross-correlation yields a velocity vector which describes the average direction and speed of particle displacement. This process is repeated for each area of interrogation. Thus, the result of a DPIV analysis is a complete velocity vector field for the selected area of interest.

Figure 1. Overview of cross-correlation analysis.

An important advantage of DPIV over more traditionally used methods of flow visualization, such as the tracking of dye fronts, is that the high particle seeding density results in a uniform distribution of velocity vectors. Accordingly, there are no gaps in the flow field and so the details of the flow can be studied precisely. Furthermore, all image processing is done computationally rather than by hand and so the analysis can be performed quickly and with improved accuracy.