Automatically adjusting cloud movement prediction model from satellite infrared images

Tracking and predicting cloud movement is one the most important step in pluviometry. The purpose of this study is to develop a self adjusting technique to predict the movement of clouds from a series of infrared (IR) images. The first task is to identify clouds from the images. This is done by clustering the images using K-means algorithm. Clouds are identified from segmented image and the large clusters are extracted. Center of Mass (CoM) is calculated for each cloud cluster and its position after every 30 minutes is predicted. Whenever the original position deviates from the predicted values, the model automatically adjusts itself with the change and the next prediction becomes closer to original values of position.