Optical-Flow Method Based On Multi-scale Used In PIV

The velocimetry instruments or methods such as ultrasonic ADV and laser LDV adopted in hydrodynamic experiments and aerodynamic experiments generally can only measure single point of flow field, and most instruments must touch and come into flow field and cannot measure two-dimensional flow field. Touching single point measurement instruments cannot fulfill the measurement requirement when flow velocity changing is sensitive to exterior disturbance and in space, and when measurement of two-dimensional velocity distribution is needed. PIV technique is a kind of untouched flow field measurement, which can get the movement information of flow field by cross-correlation analysis of two flow field images. The great promising undisturbed flow field measurement technique gets over the limitation of the touching single point measurement instruments and can measure plane two-dimensional flow field. PIV technique has already been applied to two-dimensional flow field measurement of hydrodynamic experiments and aerodynamic experiments, and provides an effective way to study velocity distribution of flow field for scientific researchers.Traditional methods used in PIV are almost based on area-correlation. This method is used diffusely for its simple concept, easily implement and ripe technology. But such method was proven to have several bugs, especially in too many points mistook, the flow field is not smooth. So there are lots of extra works to be done. Such problems improve the working amounts and descend the accuracy of the result.The paper takes movement analysis theory of computer vision into image processing into PIV. This method corrects the disadvantage in traditional method through smooth restriction term and Multi-scales technology. And it improves the quality of the vector field. Optical-flow is a method concerning computer vision abstracted from hydrodynamics. It abstracts optical-flow restriction term from the continuity equation in hydrodynamics. Through smooth restriction it gained the second restriction term and the two restriction terms were made up to be optical-flow equation. Through these two equations and iterative calculation, we can deserve the velocity of each pixel. During the computing the Pyramid data structure also called multi-scales data structure is used, we can get smooth and more accuracy field through such method. Also the speed for computing is improved. At last we will demonstrate advantages of optical-flow method through contrasting the optical-flow method and traditional method.In the process of programming, plenty of senior characteristics such as templates and inline function are used to improve the efficiency of the program.