Abstract
To perform the cloud detection and removal, a novel automatic supervised approach is proposed. Here in the cloud detection task there are mainly two sections. one is training section where the detector is designed using the feature data and ground truth of the training image. The feature data is formed by extracting and stacking different features such as colour, statistical information, texture, structure, haar like features and reflectance. Other is testing section. After training the detector the feature data of the testing image is applied to get the saliency map. After that some cloud refinement techniques are applied to get the final cloud detected result. The complemented detected output is taken as a mask and applied to the testing image. Then it is given to the cloud removal technique. Singular value decomposition is used in cloud removal. The quantitative analysis of efficiency of detection in proposed method is evaluated.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have