Real Time Reconstruction of Fluid in Video

Abstract

This paper puts forward a new method of realtime reconstruction of fluid in natural scene. It takes the measure of combination of image analysis and LBM (Lattice Boltzmann Methods). First, employs LK (Lucas-Kanade) method to calculate the dense optical flow, and then takes LBM to obtain the joint force of central particles for the initial result. After backfilling the velocity vectors field, it adopts the K-means cluster to obtain several classes, in each class, it takes advantage of the Rayleigh distribution to fit the height field of fluid. Finally, the reconstruction result of fluid is obtained. In addition, it demonstrates the results of the height field of fluid in the experiment. Further experiments shows that it is a valid method of fluid reconstruction with real time and can be used in the study of natural landscape fluid with efficiency and feasibility.