关键词: 光流法, 共轭梯度法, 自适应采样, 荧光油膜, 速度场

Abstract: The existing optical flow methods takes a long time to solve the friction measurement equations based on fluorescent oil film, which restricts the application of fluorescent oil film tests in wind tunnel. Therefore, a spatially adaptive fast optical flow of fluorescent oil film velocity field is proposed. The conjugate iterative formula of optical flow is constructed, with the smaller storage capacity, higher convergence and stability than those of Jacobian-like matrix segmentation iteration used by the existing optical flow methods; On the other hand, based on the grayscale gradients of a given fluorescent oil film image, the prominent flow structures can be quickly obtained by a spatially adaptive down sampling method, and then up sampling to obtain the refined flow structures by conjugate iteration and executing global optimization, and the speed to solve velocity field of fluorescent oil film is greatly improved. Simulation examples show that the calculation speed of this algorithm is 26.6% higher than that of the existing optical flow method under the given convergence error limits, and the average errors is reduced by 1.5%, and the maximum error is reduced by 2.3% under the same preset number of iterations. The results of the oil film tests in the wind tunnel show that the wall flow phenomenon is correct, and the flow spectrum is clear and precise, which are measured by this algorithm. Compared with the existing optical flow method, its speed is increased by 23% on average, and it has a great engineering appliance.

Key words: optical flow, conjugate gradient method, adaptive sampling, fluorescent oil film, velocity field