嫦娥二號線陣CCD影像的月表DEM提取技術(shù)研究
發(fā)布時間：2018-08-24 19:03
【摘要】：嫦娥一、二號探月衛(wèi)星的成功發(fā)射，標志著我國的深空探測技術(shù)已經(jīng)走入了世界前列。嫦娥二號衛(wèi)星搭載的CCD立體相機的空間分辨率高達7米，利用其立體像對可以生成高分辨率的月表DEM，對后續(xù)的登月計劃及數(shù)字月球平臺的搭建都具有重要意義。因此，本文圍繞著如何利用嫦娥二號的CCD立體影像數(shù)據(jù)提取月表DEM進行了系統(tǒng)的理論分析和實踐驗證。 本文對嫦娥二號衛(wèi)星所搭配的CCD立體相機的成像原理及影像數(shù)據(jù)的存儲格式進行了細致地研究，根據(jù)嫦娥二號影像數(shù)據(jù)的特點及月球遙感的特殊性，明確了本研究工作需要解決影像立體定位和影像匹配兩個難點，這是提取月表DEM的前提條件。為解決嫦娥二號CCD影像的立體定位問題，即解算外方位元素，本文系統(tǒng)地研究了線陣CCD影像的工作原理和兩種主要的傳感器模型——物理傳感器模型和通用傳感器模型，通過分析有理函數(shù)模型這一種與傳感器參數(shù)無關(guān)的外方位元素解算模型的原理及特點，并借鑒其它的搭載線陣CCD立體相機的高分辨率商業(yè)衛(wèi)星的影像定位方法，提出了基于有理函數(shù)模型的嫦娥二號CCD影像立體定位方法，在無衛(wèi)星軌道參數(shù)的情況下，成功的利用月面控制點解算了嫦娥二號衛(wèi)星CCD影像的外方位元素。由于月球影像紋理特征不明顯，而且其影像灰度受光照影響也有較明顯的差別，利用傳統(tǒng)的基于灰度的影像匹配方法對嫦娥二號影像的立體像對進行匹配難以得到可靠的匹配結(jié)果，因此本文選擇了基于特征的匹配算法——SIFT算法進行月面立體影像匹配，并進行了編程實現(xiàn)和結(jié)果分析，證明此方法是一種行之有效的月面立體影像匹配方法。最后，本文以ENVI軟件的地形分析模塊為平臺，以月球北極地區(qū)的部分影像為實驗數(shù)據(jù)，完成了該區(qū)域的DEM提取工作，生成的DEM分辨為15米，，高程誤差在30米以內(nèi)。實驗結(jié)果證明本文的方法切實可行，完全適用于嫦娥二號CCD立體影像的處理和DEM提取中，根據(jù)本文設(shè)計的方法可提取出高分辨率、高精度的全月球DEM。 本文的創(chuàng)新點有： （1）提出了一種基于有理函數(shù)模型的嫦娥二號CCD影像的立體定位方法。即利用嫦娥一號激光高度數(shù)據(jù)與嫦娥二號CCD影像控制點數(shù)據(jù)相結(jié)合的方法生成月面控制點，通過月面控制點解算出有理函數(shù)模型系數(shù)，實現(xiàn)了影像的外方位元素解算。 （2）提出了一種利用SIFT算法解決月表遙感影像匹配算法。實現(xiàn)了對嫦娥二號獲取的線陣CCD不同視角影像間的匹配及同名點的自動、批量生成，保證了提取月表DEM的精度，提高了工作效率。
[Abstract]:The successful launch of the Chang'e 1 and 2 lunar probe satellites marks that China's deep space exploration technology has entered the forefront of the world. The spatial resolution of the CCD stereo camera on the Chang 'e-2 satellite is as high as 7 meters. It is of great significance for the subsequent lunar landing plan and the construction of the digital lunar platform to use the stereo image pair to generate high-resolution DEM,. Therefore, this paper makes a systematic theoretical analysis and practical verification on how to extract the monthly table DEM from the CCD stereoscopic image data of Chang 'e-2. In this paper, the imaging principle and storage format of CCD stereo camera for Chang 'e-2 satellite are studied in detail. According to the characteristics of Chang 'e-2 image data and the particularity of lunar remote sensing, It is clear that this work needs to solve the two difficulties of image stereo positioning and image matching, which is the precondition of extracting monthly table DEM. In order to solve the problem of stereo positioning of Chang 'e-2 CCD image, that is, to solve the external azimuth element, the working principle of linear CCD image and two main sensor models, physical sensor model and universal sensor model, are systematically studied in this paper. By analyzing the principle and characteristics of the rational function model, which is an external azimuth element solution model independent of sensor parameters, and drawing lessons from other high-resolution commercial satellite image positioning methods with linear array CCD stereo camera, this paper analyzes the principle and characteristics of the rational function model, which is an external azimuth element solution model independent of sensor parameters. In this paper, a rational function model based stereo positioning method for Chang 'e-2 CCD image is proposed. In the absence of satellite orbit parameters, the outer azimuth elements of Chang 'e-2 satellite CCD image are successfully calculated by using the lunar surface control points. Because the texture feature of lunar image is not obvious, and its image grayscale is influenced by illumination, It is difficult to obtain reliable matching result by using traditional grayscale image matching method to match the stereo image of Chang 'e-2 image. Therefore, the feature based matching algorithm, sift algorithm, is selected for lunar stereo image matching in this paper. Programming and result analysis show that this method is an effective method for lunar stereo image matching. Finally, taking the terrain analysis module of ENVI software as the platform and some images of the lunar arctic region as experimental data, the paper completes the DEM extraction work in this region. The resulting DEM resolution is 15 meters and the elevation error is less than 30 meters. The experimental results show that the proposed method is feasible and suitable for the processing of Chang 'e-2 CCD stereo image and the extraction of DEM. According to the method designed in this paper, the high resolution and high precision DEM. can be extracted from the moon. The innovations of this paper are as follows: (1) A new stereo positioning method of Chang 'e-2 CCD image based on rational function model is proposed. The method of combining Chang 'e-1 laser height data with Chang 'e-2 CCD image control point data is used to generate the lunar surface control point, and the rational function model coefficient is calculated by the lunar surface control point solution. The external azimuth element solution of the image is realized. (2) A SIFT algorithm is proposed to solve the remote sensing image matching algorithm of the moon table. The matching of different angle images of linear CCD acquired by Chang 'e-2 and the automatic and batch generation of the same name points are realized. The precision of DEM extraction from monthly table is guaranteed and the working efficiency is improved.
【學(xué)位授予單位】：成都理工大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2012
【分類號】：P184;TP751
本文編號：2201742
[Abstract]:The successful launch of the Chang'e 1 and 2 lunar probe satellites marks that China's deep space exploration technology has entered the forefront of the world. The spatial resolution of the CCD stereo camera on the Chang 'e-2 satellite is as high as 7 meters. It is of great significance for the subsequent lunar landing plan and the construction of the digital lunar platform to use the stereo image pair to generate high-resolution DEM,. Therefore, this paper makes a systematic theoretical analysis and practical verification on how to extract the monthly table DEM from the CCD stereoscopic image data of Chang 'e-2. In this paper, the imaging principle and storage format of CCD stereo camera for Chang 'e-2 satellite are studied in detail. According to the characteristics of Chang 'e-2 image data and the particularity of lunar remote sensing, It is clear that this work needs to solve the two difficulties of image stereo positioning and image matching, which is the precondition of extracting monthly table DEM. In order to solve the problem of stereo positioning of Chang 'e-2 CCD image, that is, to solve the external azimuth element, the working principle of linear CCD image and two main sensor models, physical sensor model and universal sensor model, are systematically studied in this paper. By analyzing the principle and characteristics of the rational function model, which is an external azimuth element solution model independent of sensor parameters, and drawing lessons from other high-resolution commercial satellite image positioning methods with linear array CCD stereo camera, this paper analyzes the principle and characteristics of the rational function model, which is an external azimuth element solution model independent of sensor parameters. In this paper, a rational function model based stereo positioning method for Chang 'e-2 CCD image is proposed. In the absence of satellite orbit parameters, the outer azimuth elements of Chang 'e-2 satellite CCD image are successfully calculated by using the lunar surface control points. Because the texture feature of lunar image is not obvious, and its image grayscale is influenced by illumination, It is difficult to obtain reliable matching result by using traditional grayscale image matching method to match the stereo image of Chang 'e-2 image. Therefore, the feature based matching algorithm, sift algorithm, is selected for lunar stereo image matching in this paper. Programming and result analysis show that this method is an effective method for lunar stereo image matching. Finally, taking the terrain analysis module of ENVI software as the platform and some images of the lunar arctic region as experimental data, the paper completes the DEM extraction work in this region. The resulting DEM resolution is 15 meters and the elevation error is less than 30 meters. The experimental results show that the proposed method is feasible and suitable for the processing of Chang 'e-2 CCD stereo image and the extraction of DEM. According to the method designed in this paper, the high resolution and high precision DEM. can be extracted from the moon. The innovations of this paper are as follows: (1) A new stereo positioning method of Chang 'e-2 CCD image based on rational function model is proposed. The method of combining Chang 'e-1 laser height data with Chang 'e-2 CCD image control point data is used to generate the lunar surface control point, and the rational function model coefficient is calculated by the lunar surface control point solution. The external azimuth element solution of the image is realized. (2) A SIFT algorithm is proposed to solve the remote sensing image matching algorithm of the moon table. The matching of different angle images of linear CCD acquired by Chang 'e-2 and the automatic and batch generation of the same name points are realized. The precision of DEM extraction from monthly table is guaranteed and the working efficiency is improved.
【學(xué)位授予單位】：成都理工大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2012
【分類號】：P184;TP751
【參考文獻】
相關(guān)期刊論文 前1條
1 劉經(jīng)南,魏二虎,黃勁松;深空網(wǎng)及其應(yīng)用[J];武漢大學(xué)學(xué)報(信息科學(xué)版);2004年07期
相關(guān)會議論文 前1條
1 楊甲森;趙娜;;行星數(shù)據(jù)系統(tǒng)研究[A];中國空間科學(xué)學(xué)會空間探測專業(yè)委員會第十九次學(xué)術(shù)會議論文集（上冊）[C];2006年
本文編號：2201742
本文鏈接：http://sikaile.net/kejilunwen/tianwen/2201742.html
