Image Restoration Based on Statistic PSF Modeling for Improving the Astrometry of Space Debris

Abstract

Space debris is a special kind of fast-moving, near-Earth objects, and it is also considered to be an interesting topic in time-domain astronomy. Optical survey is the main technique for observing space debris, which contributes much to the studies of space environment. However, due to the motion of registered objects, image degradation is critical in optical space debris observations, as it affects the efficiency of data reduction and lowers the precision of astrometry. Therefore, the image restoration in the form of deconvolution can be applied to improve the data quality and reduction accuracy. To promote the image processing and optimize the reduction, the image degradation across the field of view is modeled statistically with principal component analysis and the efficient mean point-spread function (PSF) is derived from raw images, which is further used in the image restoration. To test the efficiency and reliability, trial observations were made for both low-Earth orbital and high-Earth orbital objects. The positions of all targets were measured using our novel approach and compared with the reference positions. The performance of image restoration employing our estimated PSF was compared with several competitive approaches. The proposed image restoration outperformed the others so that the influence of image degradation was distinctly reduced, which resulted in a higher signal-to-noise ratio and more precise astrometric measurements.

Automated clustering method for point spread function classification

“…It is worth noting that PSF classification is a clustering problem and for stars with the magnitude of 11, we do not know why the PSFs are not clustered distinctly. There are two possibilities: (i) these PSFs are nearly the same essentially, so applying the average of these PSFs for subsequent processing is a sound method (Sun & Jia 2017); (ii) the classification algorithm failed for stars with a magnitude of 11 so we should attempt to optimize the method.…”

Optimization of Exposure Time Division for Multi-object Photometry

“…All saturated stars were rejected. Using the star images across the whole field of view, we utilized the PCA-based technique presented in (Jee & Tyson 2011;Sun & Jia 2017) to obtain the shape of the stellar profile as a function of the position on the CCD. The normalized stellar profiles, I n , were also applied in the GA to calculate the binary weights ω.…”