OPTICAL-FLOW AND CROSS-CORRELATION ALGORITHMS FOR ANALYZING JET FLOW

Abstract

Data processing for obtaining global velocity fields is important in fluid mechanics. This study presents a hybrid algorithm for extracting velocity vectors from particle image velocimetry (PIV) images, which were taken during the experimental process. The hybrid method used both cross-correlation and optical-flow algorithms. Firstly, a cross-correlation algorithm was used to extract initial velocity from PIV images. The optical-flow algorithm was, then, applied for refined velocity fields with initial estimation by cross-correlation results. The proposed method was applied for recovering velocity vectors from a jet flow. It was shown that the hybrid method shows a similar pattern to the cross-correlation algorithm. Additionally, the resolution was much improved by the proposed algorithm in comparison to cross-correlation results. Both averaged and instantaneous velocity fields were illustrated in this study. The effect of the Lagrange multiplier and interaction number on the results of the hybrid method was investigated. The proposed method shows high ability in extracting velocity fields from PIV images.