马海蓉 丁飞 章华涛 张海涛 庄衡衡 张登银

摘要：高精度光柵测量系统和光谱识别定标对于大口径空间天文望远镜观测至关重要，将图像的智能识别与光栅测量系统相结合，可以解决传统光栅测试过程中目标识别困难和特征难以提取的问题。由于目标光源点光谱成像的特点以及背景噪声的干扰，图像目标的自动识别和位置提取精度受限。该文设计并构建基于观测图像识别的光栅测试分析系统，利用光栅的分光特性，结合图像传感器进行光电转换和特征识别，通过分析目标光栅图像的关键特征，利用图像预处理算法对光栅图像进行反色和模糊去噪，获得更清晰的光斑特征，通过密度质心法提取光斑中心，再由高斯曲面拟合提取光栅图像的目标中心和像素分布特征。实验结果表明，相比传统密度质心法，该文方法能够准确提取光栅图像中多目标的中心，像素幅值识别精度提升1个像素以上。

关键词：光栅测试;图像识别;密度质心法;高斯曲面拟合

中图分类号： TP391.4文献标志码： A文章编号：1674–5124（2021）12–0029–05

Research on optimization method of target location based on raster image recognition

MA Hairong1，2，DING Fei1，2，ZHANG Huatao3，4，ZHANG Haitao1，2，ZHUANG Hengheng1，2，ZHANG Dengyin1，2

（1. Jiangsu Key Laboratory of Broadband Wireless Communication and Internet of Things， Nanjing University of Postsand Telecommunications， Nanjing 210003， China;2. School of Internet of Things， Nanjing University of Posts andTelecommunications， Nanjing 210003， China;3. Nanjing Institute of Astronomical Optics &Technology，NationalAstronomical Observatories， CAS， Nanjing 210042， China;4. Key Laboratory of Astronomical Optics & Technology，CAS， Nanjing 210042， China）

Abstract： High-precision raster measurement system and spectral recognition calibration are essential for large- aperture space telescope observations. Combining intelligent image recognition with the raster measurement system can solve the problem of difficulty in target recognition and feature extraction in the traditionalraster process. Due to the characteristics of spectral imaging of target light source points and the interference ofbackground noise， the accuracy of automatic identification and location extraction of image targets is limited.This paper designs and constructs araster test and analysis system based on observation image recognition. It usesthespectralcharacteristicsof therasterandcombinestheimagesensortoperformphotoelectric conversionandfeature recognition. Byanalyzing the keyfeaturesof the targetgratingimage， theimage preprocessing algorithm is used to perform the raster image. Reverse color and blur and denoise to obtain a clearer spot feature. The spot center is extracted by the density centroid method， and then the target center and pixel distribution characteristics of the raster image are extracted by Gaussian surface fitting. Experimental results show that compared with the traditional density centroid method， this method can accurately extract the center of multiple targets in the raster image， and the pixel amplitude recognition accuracy is improved by more than 1 pixel.

Keywords： raster test; image recognition; density centroid method; Gaussian surface fitting

0引言

光栅是一种把入射光束的振幅、相位或两者同时加上一个周期性的空间调制的光学装置，具有抗干扰性强、体积小、耐腐蚀等突出优点[1]。光的波动性引起的衍射现象最早由意大利的弗朗西斯科·格里马第发现[2]。光栅测量系统常用于激光光源波段特性分析，如波长测量和光栅常数测量等，其基本原理是光源透过光栅发生色散，投影到图像传感器上实现成像，再对采集到的光栅图像进行图像识别[3]。……