Research Of Cloud Detection Based On Multi-Temporal Thermal Infrared Data

Cloud detection is one of the most important tasks and key technologies during remote sensing image processing. This paper presented a cloud detection method based on multi-temporal thermal infrared data. Compared with traditional methods, this method could not only provide a better discrimination of cloud-free and cloud-cover pixels, but also have the anti-interference ability for snow. Landsat8 data around Beijing and Hotan district were selected for checking the accuracy of this method. The validation results showed that this method could extract cloud boundary accurately, and had a good extraction effect for thin clouds. Most important of all, it could accurately distinguish cloud pixels from snow cover. Whether in flat areas or mountain areas, for thick clouds or scattered thin clouds, the overall accuracy of this method could be always kept above 95%. This showed that the proposed method would have a very broad application prospect in cloud detection of remote sensing images.