High-precision Astronomical Image Mosaics

Astronomical images taken by Long-focus telescope usually don’t have large field of view but high precision and luminosity. In order to obtain astronomical images with high precision and a large field of view, image mosaics are implemented. Several methods proposed here consist of coordinate transformation and pixel fusion. Three different transformation schemes are experimented in coordinate transformation. Respectively, the first one is based on space projection transformation of plane-to-plane. The second one is based on spherical coordinate transformation. And the last one is based on auxiliary coordinate system with a faster speed. Our study shows that the second scheme proposed here acquires composite images with highest precision and lowest speed among three schemes. Its coordinate transformation combined with stars’ theoretical position in UCAC4. Besides, the first scheme can obtain a high precision, and can be finished in least time. But it is not rigorous to some degree. However, the last one is not good at precision in our experiment, though it is rigorous one. Maybe its bad precision is caused by which some calculations are too strongly depends on point of tangency. It’s not clear whether its strict conditions in point position of tangency makes it not applicable in our research. After all the coordinate transformations are completed, the image fusion technologies are used to merge all images into one with a field of view as large as the images aggregation. In addition, this paper derives that fast plane-to-plane(the 3rd scheme) is a variation expression of eight parameters of the model.These algorithms are capable to composite a lager CCD frame which retain precision in position and has a higher SN than the original with many CCD frames. So far, our composograph is available already to find moving objects and detect faint objects. Furthermore, the method is actually tested with many CCD frames taken by the 2.4 m Telescope at Yunnan Observatory. As is shown, the method can yield a quite good CCD frame with a wide field of view when scheme two is performed. Lots of tests are done with scheme two, and it also is achieved to combine multiple frames into a lager frame which conserves precision and has even higher signal-to-noise ratio(S/N). The experiments have also shown that the star positional error introduced by the frame-combination has reduced to about 0.02 pixels since raw CCD frames are corrected for their geometry distortions by Peng et al while it is about 0.04 before geometry correction. The experiments therefore confirm the efficiency of Peng et al.’s geometric distortion solution and the necessity of correcting geometric distortions.