Underwater crowd flow detection using Lagrangian dynamics

Abstract

Crowd flow in an underwater scene is used to study fish schooling characteristics. The hydromechanics of fish schooling is quite complicated because it varies in the types of fish underwater. Fish schooling detection contributes toward the study of their locomotion, perception and behavior. For this scenario, we opted to use Lagrangian particle dynamics had been applied to examine the trajectories of an individual particle in the flow. The mathematical algorithm of the Lagrangian Coherent Structure (LCS) is the Finite-time Lyapunov Exponent (FTLE) field the divergence of the particles. The FTLE field is a boundary that divides the flow with respect to its distinct flow dynamics. The algorithm can also do real-time video analyses which can easily be implemented in Autonomous Underwater Vehicles (AUVs). The system is efficient enough to perform the detection of fish schooling as a group flow.