Advancement Through Research
The tracking and forecasting of satellite cloud images play a vital role in weather prediction by providing critical data on cloud behavior and movement. In this study, we present a novel approach for the automatic tracking of multiple cloud clusters using the VFC Snake model. This method is grounded in the principles of contour extraction and comprehensive analysis of cloud clusters, enabling the system to dynamically determine the new locations of these target cloud clusters in real time.
To enhance the accuracy of the tracking process, we have developed a specialized detection algorithm. This algorithm serves to correct and refine the tracking results generated by the Snake model, yielding more accurate contour curves that represent the boundaries of the cloud clusters with greater fidelity.
Regarding the forecasting of cloud images, we integrate the displacement data of the target cloud clusters—acquired during the tracking phase—into a cross-correlation matching process. This integration significantly improves the matching accuracy of the cross-correlation method, resulting in more precise estimates of cloud motion vectors. These vectors are essential for predicting future cloud positions and behaviors.
Our experimental results indicate that the tracking method based on contour detection and analysis not only operates at high speeds but also achieves remarkable accuracy. Additionally, we can observe the evolution of cloud clusters over time, including phenomena such as splitting, merging, dying out, and new formations. The results demonstrate a strong potential for improving forecasting accuracy, showcasing the effectiveness of our proposed method in real-world applications.
Active Contour, Satellite Cloud Images, Tracking and Forecasting
