节点文献
近红外星点目标成像系统噪声抑制方法研究
Study on Denoising Methods for Infrared Star Point Target Imaging System
【作者】 魏伟;
【导师】 刘恩海;
【作者基本信息】 中国科学院研究生院(光电技术研究所) , 信号与信息处理, 2013, 博士
【摘要】 白天星敏感器技术是星敏感器技术的一个重要发展方向。研究白天星敏感器技术是将现有星敏感器推广应用于机载、舰载、弹载等多平台的前提和基础。以星点目标为探测对象,研究近红外波段成像系统,开展白天观星试验等工作,是研究白天星敏感器技术的一种有效方案和技术途径。本文围绕近红外星点目标成像系统中的一项关键技术——噪声抑制方法展开研究。对星点目标成像前和成像后两个阶段噪声的抑制方法进行了深入的研究,对于解决白天星敏感器中的关键问题具有重要意义。在星点目标成像前,对低噪声红外成像电路设计技术进行了研究。在理论分析的基础上,进行了低噪声CCD驱动电路和噪声抑制电路的设计,完成了CCD星相机的设计和室内成像实验。研究了红外星相机的噪声抑制技术,包括低噪声驱动电路、信号获取电路和制冷电路的设计技术、相机自带两点法非均匀性校正方法等。然后设计了外场观星试验方案,搭建了试验系统,完成了白天观星试验,采集了大量的红外星图,验证了试验方案的可行性和噪声抑制措施的有效性。星点目标成像后,针对红外星图中的单点噪声,提出了基于4领域和基于列向量扫描的单点噪声抑制方法;针对红外星图中的条纹噪声,提出了列直方图偏移校正(CHOC)和列均值偏移校正(CMOC)两种算法。并用外场实拍星图完成了对上述算法噪声抑制效果的验证,结果显示,去除单点噪声和条纹噪声后,星图质量得到了较大的改善,星图的星点目标峰值信噪比提高了近2倍。为进一步提高星点背景对比度,本文研究了多种背景滤波算法,在实验分析的基础上提出了一种适用于红外星图的wiener2_top-hat组合滤波方法。用模拟理想星图与外场观测星图叠加构成合成星图的方法,对文中算法的滤波效果进行验证,分析了星点目标峰值信噪比的变化和星点质心的偏移误差。分析的结果表明:文中序列算法处理后星图的峰值信噪比提高到了原始星图的3~4倍,星点质心偏移仅为1/30像素,算法具有较高的精度。本文围绕近红外星点目标成像系统的噪声抑制方法展开研究。分析了系统中各种噪声的特性和抑制方法,提出了具有针对性的解决思路和方案,在外场试验和星图处理实验中取得了良好的效果,为进一步开展近红外白天观星试验研究奠定了基础。
【Abstract】 Daytime star sensor technology is an important development direction of starsensor technology. The research on daytime star sensor technology is the premise andfoundation of extending the application platform of present star sensor to airborne,shipborne, missileborne and other platforms. Using the stars as the detecting targets,researching infrared imaging systems and launching daytime star observationexperiments is one effective scheme and technical approach of daytime star sensortechnology research.The thesis did research around the denoising methods, which was one keytechnology of infrared star point target imaging system. The thesis made intensive studyon denoising methods in two sections: before and after star point target imaging, whichwas of great significance in resolving the key problems of daytime star sensor.Before the star point target imaging, the technology of designing low noise infraredimaging circuits was researched. On the basis of theoretical analysis, a low noise CCDdriving circuit and a denoising circuit were designed, a CCD star camera was designedand an inroom imaging experiment was accomplished. Denoising methods for infraredstar camera were researched, which contain the design technique of low noise drivingcircuit, signal acquisition circuit, cooling circuit, two-piont nonuniformity correctionmethod integrated in infrared camera and etc. Then the scheme of daytime starobservation was designed, the experimental system was set up, the experiment wasaccomplished and sevral infrared star maps were collected. Through the experiment, theeffectiveness of scheme and denoising methods were confirmed.After star point target imaging, for the single-point-noise, the denoising algorithmsbased on4neighborhood and column direction scanning were proposed. For the stripe- noise, the algorithms based on column histogram offset correction (CHOC) and colummean offset correction (CMOC) were proposed. When the algorithms were used to thestar maps collected in out-field experiment, the results showed that the quality of starmaps was improved largely and the star point target peak signal-noise-ratio (PSNR) wasraised to nearly2times after denoising single-point-noise and stripe-noise.To increase the star and background contrast ratio furtherly, sevral backgroundfiltering algorithms were analyzed. On the basis of experimental analysis, a composedfiltering algorithm wiener2_top-hat, which was suitable for infrared star map, wasproposed. We added the simulated ideal star map to the observed star map to make up acomposed star map, and use the composed star map to confirm the effectiveness ofdenoising algorithms, the star-point target PSNR and offset error of star-point centroidwere analyzed. The result showed that, after the serial denosing processing, the Psnrwas raised to3~4times of the raw star map, and the offset error of star-point centroid isonly1/30pixel, the accuracy of the algorithms in the paper is high.The thesis did an in-depth study on the denoising methods for infrared star pointtarget imaging system. The characteristics and denosing methods were analyzed, andspecific ideas and schematics were put forward, the result of out-field experiment andexperiment of processing to star map showed good results, which laid a foundation forinfrared daytime star observation.
【Key words】 daytime star sensor; spectral filtering; infrared star map; single-point-noise; stripe noise;
- 【网络出版投稿人】 中国科学院研究生院(光电技术研究所) 【网络出版年期】2014年 02期
- 【分类号】TN216
- 【下载频次】426
- 攻读期成果