改进Zernike矩的小行星接近过程图像的亚像素边缘检测方法  

Improved subpixel edge detection for asteroid approach process images with Zernike moments

作　　者：刘月鹏 张新宇[3] 乔栋[1,2] 徐义奇 李翔宇[1,2] LIU YuePeng;ZHANG XinYu;QIAO Dong;XU YiQi;LI XiangYu(School of Aerospace Engineering,Beijing Institute of Technology,Beijing 100081,China;Key Laboratory of Flight Dynamics and Control,Ministry of Education,Beijing 100081,China;Beijing Institute of Control Engineering,Beijing 100190,China)

机构地区：[1]北京理工大学宇航学院,北京100081 [2]飞行动力学与控制教育部重点实验室,北京100081 [3]北京控制工程研究所,北京100190

出　　处：《中国科学：技术科学》2024年第8期1473-1484,共12页Scientia Sinica(Technologica)

基　　金：国家自然科学基金(批准号:51827806,12002028);中国科协青年托举人才项目(编号:YESS20210296);北京理工大学青年教师学术启动计划(编号:XSQD-202101012)资助。

摘　　要：对小行星图像边缘检测是实现小天体接近过程中精确自主导航的关键,对威胁小天体防御应对措施的精准实施至关重要.现有的小行星图像分割算法精度较低,且图像中存在星点密集等问题,给后续的图像处理带来挑战.本文针对小行星接近过程中图像的亚像素目标分割问题,提出一种改进Zernike矩的亚像素边缘检测方法.首先,使用双边滤波器进行图像预处理;其次,使用改进的形态学梯度滤波算子实现像素级边缘的粗定位;然后,使用改进的Zernike矩模型建立7×7模板,并通过Otsu法计算得到Zernike矩的最佳分割阈值,实现边缘检测的亚像素精确定位;最后,使用细化算法对亚像素图像边缘进行提取.仿真结果表明,该算法可以实现对图像边缘的高精度亚像素定位,研究可以为未来小行星防御的自主导航与定轨提供参考.Edge detection of asteroid images is the key to achieving precise autonomous navigation during the approach process of asteroids,and it is crucial to implement precise defense measures against threats to asteroids.The existing asteroid image segmentation algorithms are inaccurate and have problems such as dense star points in the images,posing challenges to subsequent image processing.This article proposes an improved Zernike moment-based subpixel edge detection method for subpixel object segmentation in asteroid-approaching images.Firstly,a bilateral filter for image preprocessing was used.Secondly,an improved morphological gradient filter operator was used to achieve coarse pixel-level edge localization.Then,an improved Zernike moment model was used to establish 7×7 templates and calculate the optimal segmentation threshold of Zernike moments using the Otsu method to achieve subpixel accurate positioning of edge detection.Finally,the refinement algorithm is used to extract the edges of subpixel images.The simulation results show that this algorithm can achieve high-precision subpixel positioning of image edges,and research can provide a reference for future autonomous navigation and orbit determination of asteroid defense.

关 键 词：亚像素 边缘检测 ZERNIKE矩 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] P185.7[自动化与计算机技术—计算机科学与技术]