GPU Accelerated Direct Cross-Correlation PIV with Window Deformation

Publication Year: 

Many kinds of PIV algorithms and systems have been proposed. Although a lot of PIV processing are highly sophisticated in terms of their accuracy and spatial and temporal resolution, the general problem of PIV is its time cost to compute the vector fields from images, which often imposes specific constraints on measuring methods. In this paper, focusing on recursive direct cross-correlation PIV with window deformation, which is one of the most popular algorithms for PIV, we propose a technique to accelerate this PIV processing using Graphics Processing Unit (GPU). We show that this PIV can be computed over 100 times faster than that computed by only CPU and over 30 PIV processing in a second can be achieved in some image size. The scalability of this algorithm is also shown.

Graduate School of Frontier Science, The University of Tokyo, Chiba, Japan