卫星相机线扩散函数FWHM测量不确定度估计及其应用
发布时间：2018-10-16 19:32
【摘要】：对卫星相机在轨点扩散函数(Point Spread Function,PSF)及调制传递函数Modulation Transfer Function,MTF)进行测量具有十分重要的理论和实际工程应用的意义。卫星相机在轨状态与地面静态相比,工作环境等条件有较大的变化,因此在轨PSF/MTF不能使用静态PSF/MTF来代替,必须通过实际获得的遥感图像进行提取。点光源法、周期靶标法、脉冲法及倾斜刃边法等是典型的从遥感图像中提取成像时刻的PSF/MTF的方法。其中,倾斜刃边法对高分辨率系统的空间响应有较高的准确性,同时对靶标的要求不如点光源法严格,因而广泛应用于国内外对高分辨率遥感系统的在轨PSF/MTF测量中。然而,卫星相机只能根据刃边靶标布设的时间和位置进行检测,不能满足长期跟踪卫星相机性能与实时在轨PSF/MTF的获取的要求,而这正是卫星相机在轨PSF/MTF监测希望达成的效果。设若倾斜刃边法能够突破靶标的限制,通过合适的自然地物目标影像就能够得到较高精度的在轨MTF测量结果,就能够对卫星相机在轨MTF进行连续的检测和实时实地MTF的获取,对卫星相机长期在轨运行性能跟踪和图像实时复原有重要意义。在倾斜刃边法的基础上,对其关键步骤边缘扩散函数(Edge Spread Function,ESF)提取方法进行研究。在MSG(Modified SG)方法的基础上进行改进,提出一种自适应SG(Self-Adaptive SG Method,SASG)方法。通过仿真实验并结合在实际遥感图像中的使用效果,验证了SASG方法的有效性。对比典型的三次样条方法(Spline Method),三次样条-SG组合方法(SplineSG Method)实验结果与Spline方法基本一致,而SASG方法有更强的抗条带噪声干扰能力。此外,SASG从算法复杂程度上相比Spline与SplineSG更为简便。因此,SASG方法有一定的实用性。结合国内外对倾斜刃边法另一关键算法亚像素边缘探测算法的研究,以Fermi函数探测方法为精度最高的探测方法,完成对倾斜刃边法关键算法的优化研究。在对倾斜刃边法进行优化的基础上,通过仿真实验研究适用倾斜刃边法的地物影像需要满足的条件,从而为倾斜刃边法摆脱靶标限制打下基础。对于刃边倾角、刃边尺寸以及噪声分布等图像条件对算法准确度影响的分析,以Nyquist频率处的MTF值(NyquistMTF)、PSF的半高宽度(Full Width at Half Maximum,FWHM)等参数为统计量形成定量的分析结果。仿真实验结果表明,地物影像倾角在6°~10°时,FWHM的计算精度在±1%以内,不确定度在2%以内;Nyquist MTF的计算精度在±5%以内,不确定度在2%以内,满足对算法准确度的要求。其中,结合精度与不确定度的考虑,8°可以作为准确度最高的倾角。图像有效尺寸随倾角变化而变化,以8°倾角为基础,要求刃边高度至少20pixel左右,刃边宽度则需要尽量大的亮暗均匀区域。为保证算法精度,信噪比应该不低于40dB。研究表明,NyquistMTF对于FWHM的变化非常敏感,因此,在不要求具体数值时,可以仅根据FWHM对成像系统性能做出定性分析。此外,由于卫星侧摆角范围在0°~40°之内,大侧摆对卫星相机采样分辨率的影响不能忽略,因此,刃边边缘灰度分布将发生变化,从而对倾斜刃边法结果产生影响。通过仿真方法及实际遥感图像的应用,分析采样分辨率变化对倾斜刃边法算法准确度的影响。实验结果表明,对同一刃边地物目标,当地面采样间隔(Ground Sampling Distance,GSD)增加时,地物影像尺寸减小,倾斜刃边法FWHM估计结果下降,PSF与MTF提取值偏高,在图像复原中引产生振铃与模糊等。
[Abstract]:It is very important to measure the point Spread Function (PSF) and Modulation Transfer Function Modulation (MTF) in the orbit point spread function (MTF) of the satellite camera. When the orbit state of the satellite camera is compared with the static state of the ground, the working environment and other conditions have great changes. Therefore, the static PSF/ MTF cannot be used instead of the orbit PSF/ MTF, and the extracted remote sensing image must be extracted. The point light source method, the periodic target method, the pulse method and the inclined blade edge method are typical methods for extracting the PSF/ MTF of the imaging time from the remote sensing image. The inclined blade edge method has high accuracy for the spatial response of the high-resolution system, and the requirement of the target is not as strict as the point light source method, so it is widely applied to the measurement of the orbit PSF/ MTF of the high-resolution remote sensing system at home and abroad. However, the satellite camera can only detect the time and position of the target layout of the blade edge, and can not meet the requirement of long-term tracking satellite camera performance and real-time acquisition of the orbit PSF/ MTF, which is the effect of the satellite camera on the track PSF/ MTF monitoring. if the inclined blade edge method can break through the limitation of the target, the accurate orbit MTF measurement result can be obtained with a suitable natural figure target image, and continuous detection and real-time MTF acquisition of the satellite camera on the orbit MTF can be carried out on the satellite camera, It is of great significance to long-term track operation performance tracking and real-time image restoration of satellite camera. The Edge Spread Function (ESF) extraction method is studied on the basis of the oblique edge method. Based on the MSG (Modified SG) method, an adaptive SG (Self-Adaptive SG Method, SASG) method is proposed. The effectiveness of SASG method is verified by simulation experiment and combined use effect in real remote sensing image. Compared with the typical cubic spline method (Spline Method), the three-cubic spline-SG combination method (SplaneSG Method) experiment results are basically the same as the Spline method, while the SASG method has stronger anti-banding noise interference capability. In addition, SASG is simpler than Spline and SplineSG from the complexity of the algorithm. Therefore, SASG method has some practicability. Based on the research of the sub-pixel edge detection algorithm of another key algorithm of inclined blade edge method at home and abroad, the method of Fermi function detection is the highest precision detection method, and the optimization research on the key algorithm of inclined blade edge method is completed. On the basis of the optimization of the inclined blade edge method, the condition that the feature image needs to be satisfied is studied by the simulation experiment, thus laying the foundation for the inclined blade edge method to get rid of the target limit. For the analysis of the influence of image conditions such as edge inclination angle, blade edge size and noise distribution on the accuracy of the algorithm, quantitative analysis results are formed based on the parameters such as the MTF value at Nyquist frequency (Nyquist MTF), the half height of PSF (Full Width at Half Maximum, FWHM). The simulation results show that the accuracy of FWHM is within 1% and the uncertainty is within 2% at 6 掳 ~ 10 掳. The accuracy of the Nyquist MTF is within 5%, the uncertainty is within 2%, and the accuracy of the algorithm is satisfied. Among them, with the consideration of accuracy and uncertainty, 8 deg. can be used as the highest accuracy tilt angle. The effective size of the image changes with the change of the inclination angle. Based on the inclination angle of 8 掳, the height of the edge edge is required to be at least about 20Pixel, and the width of the edge shall be as large as possible. To ensure the accuracy of the algorithm, the signal-to-noise ratio should not be less than 40dB. The results show that the Nyquist MTF is very sensitive to FWHM's variation. Therefore, it is possible to make qualitative analysis of imaging system performance only according to FWHM when no specific values are required. In addition, since the satellite-side swing angle range is within 0 掳 ~ 40 掳, the influence of large-side pendulum on the sampling resolution of satellite camera can't be ignored. Therefore, the edge-edge gray-scale distribution will change, which will affect the results of edge-edge method. Through the simulation method and the application of the actual remote sensing image, the influence of the variation of the sampling resolution on the accuracy of the edge-edge method is analyzed. The experimental results show that when the ground sampling interval (GSD) increases, the image size decreases, the FWHM estimation result decreases, the PSF and MTF extraction value are high, and ringing and blurring are produced in image restoration.
【学位授予单位】：中国科学院研究生院(长春光学精密机械与物理研究所)
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：V445.8
本文编号：2275407
[Abstract]:It is very important to measure the point Spread Function (PSF) and Modulation Transfer Function Modulation (MTF) in the orbit point spread function (MTF) of the satellite camera. When the orbit state of the satellite camera is compared with the static state of the ground, the working environment and other conditions have great changes. Therefore, the static PSF/ MTF cannot be used instead of the orbit PSF/ MTF, and the extracted remote sensing image must be extracted. The point light source method, the periodic target method, the pulse method and the inclined blade edge method are typical methods for extracting the PSF/ MTF of the imaging time from the remote sensing image. The inclined blade edge method has high accuracy for the spatial response of the high-resolution system, and the requirement of the target is not as strict as the point light source method, so it is widely applied to the measurement of the orbit PSF/ MTF of the high-resolution remote sensing system at home and abroad. However, the satellite camera can only detect the time and position of the target layout of the blade edge, and can not meet the requirement of long-term tracking satellite camera performance and real-time acquisition of the orbit PSF/ MTF, which is the effect of the satellite camera on the track PSF/ MTF monitoring. if the inclined blade edge method can break through the limitation of the target, the accurate orbit MTF measurement result can be obtained with a suitable natural figure target image, and continuous detection and real-time MTF acquisition of the satellite camera on the orbit MTF can be carried out on the satellite camera, It is of great significance to long-term track operation performance tracking and real-time image restoration of satellite camera. The Edge Spread Function (ESF) extraction method is studied on the basis of the oblique edge method. Based on the MSG (Modified SG) method, an adaptive SG (Self-Adaptive SG Method, SASG) method is proposed. The effectiveness of SASG method is verified by simulation experiment and combined use effect in real remote sensing image. Compared with the typical cubic spline method (Spline Method), the three-cubic spline-SG combination method (SplaneSG Method) experiment results are basically the same as the Spline method, while the SASG method has stronger anti-banding noise interference capability. In addition, SASG is simpler than Spline and SplineSG from the complexity of the algorithm. Therefore, SASG method has some practicability. Based on the research of the sub-pixel edge detection algorithm of another key algorithm of inclined blade edge method at home and abroad, the method of Fermi function detection is the highest precision detection method, and the optimization research on the key algorithm of inclined blade edge method is completed. On the basis of the optimization of the inclined blade edge method, the condition that the feature image needs to be satisfied is studied by the simulation experiment, thus laying the foundation for the inclined blade edge method to get rid of the target limit. For the analysis of the influence of image conditions such as edge inclination angle, blade edge size and noise distribution on the accuracy of the algorithm, quantitative analysis results are formed based on the parameters such as the MTF value at Nyquist frequency (Nyquist MTF), the half height of PSF (Full Width at Half Maximum, FWHM). The simulation results show that the accuracy of FWHM is within 1% and the uncertainty is within 2% at 6 掳 ~ 10 掳. The accuracy of the Nyquist MTF is within 5%, the uncertainty is within 2%, and the accuracy of the algorithm is satisfied. Among them, with the consideration of accuracy and uncertainty, 8 deg. can be used as the highest accuracy tilt angle. The effective size of the image changes with the change of the inclination angle. Based on the inclination angle of 8 掳, the height of the edge edge is required to be at least about 20Pixel, and the width of the edge shall be as large as possible. To ensure the accuracy of the algorithm, the signal-to-noise ratio should not be less than 40dB. The results show that the Nyquist MTF is very sensitive to FWHM's variation. Therefore, it is possible to make qualitative analysis of imaging system performance only according to FWHM when no specific values are required. In addition, since the satellite-side swing angle range is within 0 掳 ~ 40 掳, the influence of large-side pendulum on the sampling resolution of satellite camera can't be ignored. Therefore, the edge-edge gray-scale distribution will change, which will affect the results of edge-edge method. Through the simulation method and the application of the actual remote sensing image, the influence of the variation of the sampling resolution on the accuracy of the edge-edge method is analyzed. The experimental results show that when the ground sampling interval (GSD) increases, the image size decreases, the FWHM estimation result decreases, the PSF and MTF extraction value are high, and ringing and blurring are produced in image restoration.
【学位授予单位】：中国科学院研究生院(长春光学精密机械与物理研究所)
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：V445.8
【相似文献】
相关期刊论文 前2条
1 李剑白;;评价成象质量的光学传递函数方法(二)[J];仪器制造;1975年06期
2 ;[J];;年期
相关会议论文 前1条
1 吴帅;迟泽英;游明俊;陈文建;;光纤传像束的光学传函测量[A];全国第十次光纤通信暨第十一届集成光学学术会议（OFCIO’2001）论文集[C];2001年
相关硕士学位论文 前2条
1 陈起行;卫星相机线扩散函数FWHM测量不确定度估计及其应用[D];中国科学院研究生院(长春光学精密机械与物理研究所);2015年
2 陈圆圆;透镜测量仪的光学系统设计及MTF算法的研究与实现[D];哈尔滨工业大学;2007年
，本文编号：2275407
本文链接：https://www.wllwen.com/kejilunwen/hangkongsky/2275407.html
