Tracking hurricane paths
The South East coastal region experiences hurricane threat for almost six months in every year. To improve the accuracy of hurricane forecasts, meteorologists would need the storm paths of both the present and the past. A hurricane path can be established if we could identify the correct position of the storm at different times right from its birth to the end. We propose a method based on both spatial and temporal image correlations to locate the position of a storm from satellite images. During the hurricane season, the satellite images of the Atlantic ocean near the equator are examined for the hurricane presence. This is accomplished in two steps. In the first step, only segments with more than a particular value of cloud cover are selected for analysis. Next, we apply image processing algorithms to test the presence of a hurricane eye in the segment. If the eye is found, the coordinate of the eye is recorded along with the time stamp of the segment. If the eye is not found, we examine adjacent segments for the existence of hurricane eye. It is probable that more than one hurricane eye could be found from different segments of the same period. Hence, the above process is repeated till the entire potential area for hurricane birth is exhausted. The subsequent/previous position of each hurricane eye will be searched in the appropriate adjacent segments of the next/previous period to mark the hurricane path. The temporal coherence and spatial coherence of the images are taken into account by our scheme in determining the segments and the associated periods required for analysis.