基于时空图像频谱的时均流场重建方法

尽管大尺度粒子图像测速技术(LSPIV)在水流测速方面具有优势,但该方法得到的瞬时流场矢量正确率低,导致了时均流场重建误差大。提出了一种基于时空图像频谱的时均流场重建方法。首先利用示踪物在三维时空域中分布的相关性,通过在图像序列中设置的测速线合成时空图像;然后基于傅里叶变换的自配准性质建立时空图像纹理主方向和频谱主方向间的约束关系;最后采用由粗到精的搜索策略检测频谱主方向并估计测速线上一维运动矢量的大小。实验表明:采用此方法检测纹理图像频谱主方向可以达到0.1°的检测精度,对现场成像暗噪声类的干扰模式具有抗噪性,有效提高了时均流场重建的精确性。

Time averaged flow field reconstruction method based on spectrum of spatio temporal image spectrum of spatio temporal image

Despite that the Large Scale Particle Image Velocimetry (LSPIV) technology has advantages in detecting flow speed, the correct rate of the instantaneous velocity vector obtained using this method is low, which results in big error in the reconstruction of time averaged flow field. A time averaged flow field reconstruction method based on the spectrum of spatio temporal image is proposed. Firstly, according to the correlation of tracer distribution in 3 D spatio temporal domain, a group of spatio temporal images are generated via the velocity measuring lines set in the image sequence. Secondly, the constraint relation between the main directions of the texture and the spectrum of the spatio temporal image is built based on the auto registration property of Fourier Transform. Finally, the main direction of the spectrum is detected with a coarse to fine searching strategy and the 1 D motion vector of the velocity measuring line is estimated. The experiment results show that the main direction of the spectrum of a texture image can be detected with the accuracy of using this method. The proposed method has good noise resisting capability to the interference pattern of dark field imaging noise, and efficiently improves the reconstruction accuracy of time averaged flow field.