A Real-Time Image Processing System for Detecting and Removing Cloud Shadows on Hα Full-Disk Solar

Sky clouds lead to Hα full-disk solar observed images covered with shadows, which obscure the details of solar features. In this paper, a real-time image restoring system is introduced. For detecting and removing cloud shadows, and displaying the restoring image in real-time, Graphic Processing Unit (GPU) with Compute Unified Device Architecture (CUDA) parallel programming environment is employed to accelerate computing in the system. The parallel computing steps include:(1) identifying heavy cloud covered images by calculating the ratio of major and minor axes of a fitted ellipse; (2) recognizing recoverable cloud-covered images by calculating the symmetric of limb darkening curve; (3) removing cloud shadows by Butterworth low-pass filter in frequency domain. The total processing time is about 0.7s for one cloud-covered image; it meets the need of displaying in real-time. The processing time of each operation taken in GPU is measured. The results show that Fourier transform, inverse Fourier transform and Butterworth low-pass filtering are the most time-consuming part and together they take up 52.9% of the total time. This paper also evaluate the quality of restored image, and the conclusion is that the algorithm can effectively eliminate cloud shadows; meanwhile, it has little effect on the details of solar features. The existing problem and probable improvement are also discussed at the end of this paper.