A comparison of subpixel edge detection and correlation algorithms for the measurement of sprays

Optical diagnostic techniques are commonly used to observe the breakup of dense sprays. In order to extract quantitative data from such images, edge detection algorithms have commonly been used. However, correlation image velocimetry techniques are now also becoming available for such applications. An empirical comparison between these two techniques is demonstrated for the high-speed velocimetry of the breakup of an annular air-assisted spray. A threshold based sub-pixel interpolating edge detection algorithm is employed. Both real and synthetic images are used to determine the sensitivity of the error in these techniques to changes in both image noise and defocus, the two leading causes of information loss. It is demonstrated that correlation image velocimetry techniques are generally superior in precision and accuracy as compared to edge detection techniques for the application of spray velocimetry within a reasonable parameter space of noise and defocus.