Hybrid Optical-Flow-Cross-Correlation Method for Particle Image Velocimetry

Abstract

A hybrid method for particle image velocimetry (PIV) is developed to overcome the limitations of the optical flow method applied to PIV images with large displacements. The main elements of the hybrid method include a cross-correlation scheme for initial estimation, a shifting scheme for generating a shifted image, and an optical flow scheme for obtaining a refined high-resolution velocity field. In addition, a preprocessing scheme is used for correcting the illumination intensity change. The accuracy of the hybrid method is evaluated through simulations in a parametric space in comparison with the typical correlation methods and optical flow method. Further quantitative comparisons are made in PIV measurements in a circular air jet.

OpenOpticalFlow_PIV: An Open Source Program Integrating Optical Flow Method with Cross- Correlation Method for Particle Image Velocimetry

“…Optical flow computation sensitivity to the Lagrange multiplier is not large. The hybrid method is described in [20,21], which is recapitulated for the purpose of application.…”

“…Since the coarse-grained velocity field obtained using the crosscorrelation method has a much lower spatial resolution, data interpolation is used to generate a high-resolution velocity field with the same size as the original images. For the subpixel velocity components δu x and δu y in the image plane, they follow the equation [20]. Integration of this equation from 0 to 1 (one unit time) gives the intensity variation associated with the subpixel motion δI = -∆ X (Iδu X )-∆ y (Iδu Y ) where ∆ x are ∆ y the differences over one pixel in the x and y coordinates in the image plane, and the spatial differences are ∆x = ∆y = 1 pixel.…”

“…The dependencies of the relative velocity error of the hybrid method have been evaluated in simulations in a parametric space composed of the particle displacement, particle velocity gradient, particle image density, and particle image diameter. It is found that the hybrid method consistently performs better than typical correlation methods [20,21]. The performance of the hybrid method has been further examined in the air jet experiment by evaluating the statistical quantities such as the ensemble-averaged velocity magnitude, vorticity, turbulent kinetic energy, and Reynolds stress.…”

“…Figure 3 shows the fields of the velocity vector, velocity magnitude and vorticity extracted from the particle images, where the streamlines are also shown and the The accuracy of the hybrid method has been studied through simulations of the Oseen vortex pair in in a uniform flow comparison with the optical flow method and cross-correlation methods [20]. As indicated before, the error depends on the four parameters: particle displacement, particle velocity gradient, image density, and particle image diameter.…”

Two-Dimensional Flow Field Measurement Method for Sediment-Laden Flow Based on Optical Flow Algorithm

“…In recent years, particle image velocimetry algorithms based on the optical flow method have received increasing attention from researchers [11,12]. The optical flow method, which considers the motion vector field as a whole, is suitable for the calculation and analysis of image sequences with highly concentrated particles and continuous deformation.…”