Motion Blurred Star Image Restoration Based on MEMS Gyroscope Aid and Blur Kernel Correction

Abstract

Under dynamic conditions, motion blur is introduced to star images obtained by a star sensor. Motion blur affects the accuracy of the star centroid extraction and the identification of stars, further reducing the performance of the star sensor. In this paper, a star image restoration algorithm is investigated to reduce the effect of motion blur on the star image. The algorithm includes a blur kernel calculation aided by a MEMS gyroscope, blur kernel correction based on the structure of the star strip, and a star image reconstruction method based on scaled gradient projection (SGP). Firstly, the motion trajectory of the star spot is deduced, aided by a MEMS gyroscope. Moreover, the initial blur kernel is calculated by using the motion trajectory. Then, the structure information star strip is extracted by Delaunay triangulation. Based on the structure information, a blur kernel correction method is presented by utilizing the preconditioned conjugate gradient interior point algorithm to reduce the influence of bias and installation deviation of the gyroscope on the blur kernel. Furthermore, a speed-up image reconstruction method based on SGP is presented for time-saving. Simulated experiment results demonstrate that both the blur kernel determination and star image reconstruction methods are effective. A real star image experiment shows that the accuracy of the star centroid extraction and the number of identified stars increase after restoration by the proposed algorithm.

Blind Deblurring Using Space Target Features

“…However, the estimation error will increase as the motion distance increases. The error of the proposed method is significantly smaller than that of the algorithms in [21], [38] and [26], especially when the motion blurred length and motion blurred angle are large. When the motion blurred length is fixed, motion blurred angle error is within 0 • to 2 • ; when the motion blurred angle is fixed, the error range of the motion blurred length is between 0 pixel and 1 pixel.…”

“…In the simulation experiments, in order to meet the requirements of the space targets detection accuracy simulation, and the exposure time T of the star sensor is 1.5s, the imaging area array size is 512 pixels × 512 pixels, the lens focal length is 60 cm, the star sensitivity is 5 Mv, and the rotational angular velocity is 10 • / s. The motion blurred star image used in this part is simulated under the noise with zero mean and mean square error of 0.1. The traditional Radon transform [21], the traditional Radon transform combined with different denoising algorithm, the method in [38], the method in [26] and the method in this paper are applied to the estimation of motion parameters respectively.…”

“…For existing algorithms, the PSF of the blurred star image can be obtained by the gyroscope and satellite attitude information [2], [19], [21], [23], [24] or estimated by isolated star point [25]. Wang et al [26] propose a method to calculate blur kernel, aided by a MEMS gyroscope. To improve the performance of the star sensor under dynamic conditions, Liu et al [27] propose a gyroscope-assisted star image prediction method and an improved Richardson-Lucy (RL) algorithm based on the ensemble back-propagation neural network (EBPNN).…”

Restoration Method of a Blurred Star Image for a Star Sensor Under Dynamic Conditions

“…The original star image contains a significant amount of additive noise, which includes salt and pepper noise, Gauss white noise, and Poisson noise [11]. Additive noise is unrelated to the original information of the image, but destroys the image signal by superposition.…”

Star Image Prediction and Restoration under Dynamic Conditions

“…The motion blur of the star image is another important reason that affects the dynamic performance of the star sensor. To improve the dynamic performance of the star sensor, many scholars have done a lot of research in the field of image processing, especially on the star image deblurring algorithms [23]. According to whether the point spread function (PSF) is known or not, the deblurring methods can be classified into two typical forms: Blind image deblurring (BID) with unknown PSF, and non-blind image deblurring (NBID) with known PSF [24].…”