A computer vision approach to airflow analysis

This paper describes a computer vision system for analysing air flows. The main idea of the system is to use a trace gas to visualize the air flow and to determine the air flow by analysing the trace gas and its movement via computer vision and neural network techniques. The analysis process is based on three major stages: first, an image pair series is taken from the traced air flow; second, the 3-D shapes of trace gas objects at each image pair instant are recovered; and third, the air flow velocity field is calculated via the Self-Organizing Map from the recovered 3-D surface models of the trace gas objects. The system can be used to measure air flow profiles, to determine air flow velocities and volume fluxes, and to estimate how small (e.g. dust) particles move within the flow.