The existence of noise and interference results in the occurrence of abnormal data such as mosaic and losing signals in the receiving process of satellite cloud pictures. Relying upon artificial detection and positioning, the work efficiency and the accuracy cannot be guaranteed. An abnormal data detection method of satellite cloud pictures is studied in this paper based on Hough Transform, which improved the direction of the edge detection operator to aim at the problem of the high requirement of realtime satellite cloud pictures and largeamount calculation of traditional Hough Transform combining with the characteristics of abnormal data. Experiment results show that the improved algorithm improved the arithmetic speed by 30%. It realizes the accurate positioning and rapid detection of satellite cloud picture abnormal data and has great significance for the interference source investigation and the improvement of the reception quality of satellite cloud images in the future.