Blind Deblurring Using Space Target Features

Abstract

The star image suffers inevitably from degradation due to the high-speed motion of the space target and the long exposure time of the camera, therefore the attitude information of the star is hard to accurately obtain. This paper proposes a blind deblurring algorithm that combines shape features of space targets. First, the astronomical image is preprocessed using saliency detection. Then, considering the shape characteristics of space targets, the Minimum Bounding Rectangle (MBR) is introduced to describe the space targets. Next, the parameters of the MBR are used to estimate the point spread function (PSF). Finally, a regularization method is employed to recover the astronomical images. Experimental results on both simulation and real star images demonstrate that the proposed method reduces the error of point spread function estimation and decreases the error rate of identified space targets. At the same time, the accuracy of the star centroid extraction improves 0.2786 on average and the error of the star centroid location extraction reduces 0.059 comparing to the state-of-art method. The proposed method is of great significance for positioning, recognition and attitude determination of space targets. INDEX TERMS Blind deblurring, minimum bounding rectangle, point spread function, star image restoration.

Restoration of motion-blurred star image based on improved MIMO-UNet

“…Under the combined action of noise, a degraded fuzzy star image is formed. The degradation process of the star image [8] is represented as follows:…”

A blurred star image restoration method based on gyroscope data and enhanced sparse model

“…However, this method was sensitive to noise. Liu et al [7] introduced the minimum bounding rectangle to describe the space targets and estimate the PSF. The method was not applicable to small targets such as star points.…”