Subpixel centroid position error analysis in particle tracking velocimetry induced by the CCD pixel binning

Particle tracking velocimetry (PTV) is a non-invasive, full field optical measuring technique that has become one of the dominant tools for obtaining velocity information in fluid motion. In PTV experiments, the fluid of interest is seeded with fluorescent tracer particles, where measurement of individual particle displacements, recorded by means of digital camera at two instances of time, is further used to ascertain overall flow motion. Upper limit of a flow speed a PTV system can measure is bound by the frame rate of a camera used, and the system's accuracy is limited by the accuracy of the particle centroid estimation. In order to increase the upper limit, we investigated the use of CCD binning option, which doubles camera frame rate, preserves effective field of view, suppresses photon and readout noise of CCD at the expense of loss in spatial resolution. This study provides quantitative assessment of tradeoff between aforementioned advantages of binning over the loss in spatial resolution, which can increase uncertainty in particle centroid estimation. We carried our experiments using scientific grade PixeFly camera and analyzed 1μm and 1.9μm size red fluorescence polystyrene microspheres, placed on a quartz glass plate, by the Hirox variable zoom lens (1-10x) conjugated with a OL- 700II objective. For each binning mode (horizontal, vertical and composite), we investigated and reported estimation errors of various cross-correlation and center of mass based centroid localization methods, using more than 100.000 particle image pairs. We found that, performing vertical binning in the context of laminar flows doubled measurable flow rate, while caused only a negligible estimation error in the order of hundredth of a pixel.