Partial Correlation Algorithm Applied Research And Parallel Implementation Of The Solar Magnetic Field Observations

Image Registration Technology is an important part of Image Processing. In the process of observation of solar magnetic field, the sun itself enters into the telescope and the effective amount of light related to the magnetic field is very limited. In order to improve SNR(Signal to Noise Ratio), the captured images must be overlapped. However, because of atmospheric agitation and telescope accuracy, there are errors during overlapping all images. Therefore, Image Registration Technology is needed to achieve image matching for correcting errors. At the same time, as the large area array CCD(Charge Coupled Devices) widely used in astronomical equipments, a huge amount of data has produced. How to improve computing efficiency, process large-scale data in a short period of time and implement real-time observation? These are problems astronomers want to be solved as soon as possible.In this paper, the main works are as follows:First, there are two traditional methods in Image Registration Techniques: Cross-correlation method and Phase correlation method. In this paper, two methods have been compared, and we conclude that:Cross-correlation method is more suitable for Image Processing of35cm solar magnetic field telescope in National Astronomical Observatory Huairou Solar Observing Station(HSOS).Second, through statistical analysis of solar magnetic field images which obtained by two image registration methods, the comparison result of two Image Registration method has been evaluated. Finally it confirms that Cross-correlation method is more suitable for Image Processing of35cm solar magnetic field telescope in National Astronomical Observatory Huairou Solar Observing Station(HSOS).Third, in order to reduce the processing time, several different local areas of image are selected as a baseline window. The conclusion is that:during correlation, when the interesting area is selected as a baseline window, the quality of the interesting area in the solar magnetic field images is higher.Last, GPU has more processing power than CPU in large-scale data processing. Therefore, we improved the model of image correlation algorithm of telescope in National Astronomical Observatory Huairou Solar Observing Station(HSOS). We have implemented the parallel computing for the processing of correlation-shift overlay in solar magnetic field observation, and processing efficiency has been greatly improved.The research work of this paper provides theory basis for implement of correlation algorithm and selection of base window in correlation processing in solar magnetic field observation in National Astronomical Observatory Huairou Solar Observing Station(HSOS), it has certain significance for improving solar magnetic field observation accuracy.