Clouds and shadows have a significant impact on the visual interpretation and classification of remote sensing images.Due to their masking effect,they make it difficult to accurately identify objects in the areas they cover.Additionally,clouds can also affect temperature inversion in thermal infrared remote sensing applications.Therefore,identifying clouds and shadows is a necessary prerequisite for remote sensing applications.This paper combines strict empirical threshold filtering with local dynamic thresholding to accurately extract the spatial morphology and boundary details of cloud and shadow objects.It solves the problem of excessive false detections and missed detections caused by using fixed thresholds in conventional cloud and shadow detection methods.The proposed approach provides technical support for cloud and shadow detection in domestic high-resolution satellite data.
维普
万方数据