Windowing and deformation of PIV images for the investigation of flows with large velocity gradients

A procedure for analyzing images obtained in Particle Image Velocimetry (PIV) is presented. It consists of windowing and deformation of the images to improve the measurements in large velocity gradient flow regions; the procedure is applied recursively together as the one used to determine the optimal window offset. Tests are performed on partially synthetic images (obtained by skewing the second image of a real PIV image by a given gradient) and on images acquired in the near field of a water jet. In both conditions the new algorithm substantially improve the results obtained with the standard procedure (about 20% - 30% in the evaluation of the velocity gradient). The product of adjacent correlation table (Hart method) is also considered; on the average,although the noise effects are smoothed, the method seems that must be used with care whenlarge velocity gradients are present.