Accuracy enhancement of navigation images using blind restoration method

Abstract

Optical navigation is one of the most promising technologies in the area of deep space autonomous navigation. However, since the optical images are strict motion blurred images, it is difficult to extract the lines of sight (LOS) to the beacons to reckon the spacecraft attitude and orbital position during the deep space cruise phase. This paper proposes a new blind restoration approach to effectively recover the clear image. We use a modified median filter to eliminate the black and white noises, and also, construct a blind estimation model, built on the sparsity of the gradient of navigation image, to estimate the global point spread function (PSF). Moreover, we select a few bright beacons to recover the motion blurred image, where the average value of the beacon centroid is adopted to calculate the relative position for the optical navigation. We present the simulation and actual image restoration experiment to demonstrate the accuracy and consistency of the relative position of the recovered navigation image of the proposed method. We show that the proposed method presents superior performance in comparison with the multiple cross correlation method. The estimated PSF is close to real PSF and the distributed energy of the beacon is concentrated ensuring a high SNR, where the accuracy of the relative position is higher than 0.1 pixel.