Application Of Random Sample Consistency Algorithm In Astronomical Images

In our research group’s cluster CCD image warping process,after processing such as preprocessing,star searching,centering,and matching,a data file will be generated.And subsequent calculations will be based on the data file.In most cases,the star’s measurement coordinates in the data file have a good correspondence with its position in the reference star list.That is,the residuals of the star position measurement are very small.Then the stars can do subsequent calculations.But sometimes,after fitting the coordinates of the astrometric coordinates in the data file to the theoretical position,the residuals of the astrological coordinates appear to be disorganized,and the residuals of most astrology are very large.At this time we think that the image data is not good and may need to be removed.However,in fact,there are only a few stars having a large residual in the bad astronomical image data files,and they affect other stars in the data file,making the star’s residuals in the entire data file appear disorganized.When these stars are removed from the astronomical image data file,the remaining data files become good files.According to Fischler and Bolles’description of the random sample consistency algorithm^[1],this paper adopts the improved random sample consensus algorithm to perform simulated experimental processing on the observed image data files of the three telescopes.The film constant model for the stars saved after processing was calculated by least square method.By comparing the residuals between each star and the model,it was found that the random sample consistency algorithm can effectively eliminate the bad stars in the astronomical image data files.Specifically,first,the astronomical image is processed with image processing software for preprocessing,star searching,centering,and matching to generate a data file.Then,a simulation study was performed on the measurement coordinates（x,y）of some stars in the data file.That is,the number of simulation stars and the coordinate offsets are randomly generated,making these simulation stars become bad stars and affecting the entire simulation data file.Finally,the random sample consistency algorithm is used to process the simulation data file,eliminating bad stars and keeping good stars,so as to eliminate the bad stars in the data file.The results of processing the data files of the three telescopes using the above method show that when the proportion of bad stars in the data file is within a certain range,the use of a random sample consistency algorithm can well remove all the bad stars in the data file.Specifically,when the proportion of bad stars in the 1m telescope observation image data file is within 25%,and the proportion of bad stars in the 2m4 telescope and Bok telescope observation image data file is within 20%,random sample consensus algorithm can well remove these bad stars.