Research On Fluid Motion Estimation Method Based On Optical Flow Method

Fluid motion estimation technology plays a very important role in experimental fluid mechanics,meteorology,computational vision and so on.Analyzing the velocity vector field obtained from fluid image sequences will help researchers to have a deeper understanding of complex fluid motion.Currently,there are two main fluid motion estimation algorithms commonly used: cross-correlation method and optical flow method,both of which are easily affected by imaging conditions and cannot adapt to complex fluid motion conditions.The cross-correlation method can handle particle motions with large displacements,but the resolution of the estimation results is low.The optical flow method can ensure the accuracy and hi gh resolution of the results,but it is only suitable for solving small displacement motions.Therefore,it is of great significance to study a large displacement fluid motion estimation algorithm that is suitable for different imaging conditions and has high accuracy.In this paper,we carried out in-depth research on the large-displacement motion and illumination changes existing in the current fluid mo tion estimation technology,and propose a fluid motion estimation method based on the optical flow method.Firstly,aiming at the large-displacement motion in the image sequence,this paper proposes a new multi-resolution pyramid optical flow algorithm framework.The framework uses the cross-correlation method for a priori estimation to solve the limitations of the traditional pyramid optical flow method and compensate for large displacements.Secondly,due to changes in laser or external lighting condition s,the assumption of brightness constancy may not be guaranteed.In this paper,a linear lighting model is used to improve the robustness of the algorithm under various lighting conditions.Finally,synthetic particle images and real fluid image sequences are used to compare the proposed algorithm and other representative motion estimation algorithms.The experimental results show that the algorithm in this paper has high accuracy under complex conditions such as large displacements and light changes.In addition,two types of medical image sequences of liver and brain are selected to test the proposed algorithm.The test results show that the two-dimensional velocity vector field estimated by the algorithm in this paper can clearly show the morphological area and movement trend of liver and brain tissue.