An Automatic Processing Method for All-sky Images

In the astronomy field, the estimation of cloud amount has an important influence on telescope observation.To monitor the cloud amount, it is a widely method to use all-sky cameras to capture cloud images. At present, the estimation of cloud amount is still completely by man power, which is very time-consuming and inaccurate; and the process of discrimination is entirely dependent on personal experience. Therefore, this paper presents an automatic processing method for all-sky images. The method first uses time segmentation and difference method to remove the moon influence regions in the cloud images respectively. Then the cloud and background are segmented by threshold method of the cloudy image after the removal of the affected regions of the moon. The clustering algorithm based on gray value is used to quantify the thickness of clouds in partly cloudy images. Finally, the total cloud amount is calculated and the cloud images are automatically classified according to the method of interpreting the all-sky camera cloud images based on TMT (Thirty Meters Telescope). The experimental results show that this method can greatly improve the efficiency of cloud images interpretation, while effectively saves man power; it also achieves a high recognition accuracy of 76.67%.