Calculation of Fluid Motion Vector Using Feature Vector Clustering

In order to obtain two-dimensional motion vector from two frames of video,a method of calculating fluid motion vector based on feature vector clustering is proposed in this study.First,fluid motion vector is initialized using correlation calculation of principal component matching.Then,it analyzes the initialized results with feature vector-based clustering to obtain the main categories of the motion vector.At last,it uses the linear interpolation method of minimum distance to process the motion vectors in the main categories and obtain the dense fluid motion vector.Experimental results show that the method has the characteristics of convenience and effectiveness.It can be used in the later study of reconstruction and interactive of fluid scene.