Skip to Main Content
Investigating the characteristics of mesoscale convective clouds based on satellite infrared images is very important for strong precipitation forecast. This paper proposes a new method for cloud identification, tracking and characterization using time-varying satellite infrared cloud image sequences. After clouds are identified using image processing techniques, a couple of features are extracted as their representations, based on which feature correspondences of different clouds in consecutive images are computed to track the same cloud. Then, clouds are finally characterized according to their convective activities. This automatic cloud-tracking alternative is applied to investigate the characteristics and moving tracks of the mesoscale convective clouds that move out of the Tibetan Plateau. Experimental results are also provided to demonstrate its capability to perform correct tracking in different situations.
Systems, Man and Cybernetics, 2004 IEEE International Conference on (Volume:4 )
Date of Conference: 10-13 Oct. 2004