Study On Evaluation Method Of Objective Quality Of Sunlight Ball Short Exposure Image

Solar photosphere high resolution image is of great significance to research on the solar. In order to reduce the negative effect brought by the atmospheric turbulence on imaging quality and access high resolution image with good quality, in the process of the astronomical observation, we often adopt continuous short-exposure method to obtain a series of short exposure image, from which a better quality of selected short-exposure images are selected for reconstruction. Therefore, proper short exposure image quality evaluation index is of great significance for the image selection and subsequent image reconstructionBased on the analysis of image characteristics from a large amount of solar photosphere short exposure images, we put forward an assessment method based on median filtering gradient structure similarity (MFGS) to objectively evaluate the quality of solar photosphere short-exposure image. First, reference image is obtained through median filtering on the original image. Second, the gradient of the original image and the reference image are calculated respectively. Finally, the similarity between the gradients of the original image and the reference image is calculated. The similarity is more close to1, the image quality is better and the more close to0, the image quality is poorer.Using the solar high resolution mages samples provided by the lm New Vacuum Solar Telescope of Yunnan observatory, Chinese academy of sciences, we made experiments to compare the MFGS method, RMS method, Entropy information method and Fisher method. The experimental results show that:MFGS method has a better consistency with subjective assessment and it can effectively to assess solar photosphere short-exposure image quality; MFGS method can effectively distinguish low SNR images captured under cloud cover cases, while the other three methods are indistinguishable for this type of image; in the aspect of image content relevance, MFGS method showed a weaker correlation in image content than the other three methods. The experimental results also show that, for full disk solar images whose structures are not obvious, the assessment results of the MFGS method are not satisfactory.