Algorithm for estimating star centroids in blurred space images

Dynamic shooting conditions create the effect of blurring images, which leads to distortion of the recorded scene. To ensure the necessary efficiency of the star sensor, additional image processing methods aimed at compensating for distorting factors are required. An algorithm for estimating the centroids of stars in blurred images based on the methods of noise reduction and image deconvolution is proposed. At the noise reduction stage, a bilateral filter is applied to the original image, which makes it possible to increase the signal-to-noise ratio of the stars several times. Next, background noise is removed to detect suspected stars. The image cleared of the background is sent to the input of the Lucy—Richardson algorithm for solving the deconvolution problem. A single-pass clustering algorithm is applied to the image recovered during deconvolution to estimate the centroids of the detected stars. According to the experimental results of testing the developed algorithm on three test sets of images with different blur parameters, the described approach provides higher accuracy and reliability with fixed blur parameters than two other previously known analogues. However, to increase the efficiency of the proposed algorithm, an accurate assessment of the blur parameters is required, otherwise large errors in the estimation of coordinates are possible.