PROJECT TITLE :

Scene Learning for Cloud Detection on Remote-Sensing Images

ABSTRACT:

Cloud detection plays a significant role for remote-sensing image processing. To accomplish the task, a novel automatic supervised approach primarily based on the “scene-learning” theme is proposed in this paper. Scene learning aims at coaching and applying a cloud detector on the entire image scenes. The cloud detector herein may be a special classifier that is used to separate clouds from the backgrounds. Concretely, scene learning regards each pixel of scenes in training image as a sample, and uses it to train a cloud detector. Accordingly, the detecting method is additionally implemented on each pixel of testing image using the trained detector. Generally, scene-learning scheme contains two modules: feature data simulating and cloud detector learning and applying. We first simulate a kind of cubic structural data (conjointly named feature information) by stacking different fundamental image features, together with color, statistical data, texture, and structure. Such information synthesize different image features, and it's used for cloud detector coaching and applying. Cloud detector is meant primarily based on minimizing the residual error between the feature information and its labels. The detector is straightforward to be trained as a result of of its closed-type. Applying the detector and some necessary cloud refinement methods to the testing pictures, we tend to may finally detect clouds. We have a tendency to conjointly theoretically analyze the influence of feature number and prove that additional features lead to better performance of scene learning below sure circumstance. Comparisons of qualitative and quantitative analyses of the experimental results are implemented. Results indicate the efficacy of the proposed methodology.