An Adaptive Extracting Method for Motion-blurred Stars

In this paper, a novel method for extracting difficult to determine a suitable global threshold, and a large motion-blurred stars in high dynamic imaging environment amount of noise causes the connected domain analysis to take affected by stray light is proposed. According to the characteristic of the background noise, an adaptive threshold estimation algorithm for each pixel with down sampling method and block weighted least squares fitting is presented. Furthermore, based on the scanning method and DBSCAN clustering, a new algorithm for extracting pixels of each star is also proposed. The experimental results indicate that the proposed method can improve the accuracy and real-time performance of star image processing even affected by inhomogeneous stray light under high dynamic conditions. The centroid error can be reduced to less than 1 pixel even if the star is motion-blurred to more than 30 pixels and the peak gray value is relatively low.