發(fā)布時(shí)間：2019-01-02 07:25

【摘要】：遙感圖像是指從遠(yuǎn)距離平臺(tái)上利用光電成像載荷獲取的地物目標(biāo)圖像,常見的平臺(tái)如飛機(jī)和衛(wèi)星。由于其覆蓋范圍廣,光譜頻段豐富,在軍事和民用上都起到了巨大的作用。但是隨著人類活動(dòng)的區(qū)域日益廣泛,對(duì)于遙感圖像的整體要求也越來越高,希望在獲得高分辨率圖像的同時(shí),其覆蓋的區(qū)域更加的寬廣,以至于能對(duì)一個(gè)較大的區(qū)域進(jìn)行細(xì)致深入的分析研究,為后續(xù)的人類決策提供支持。但是,受限于目前的技術(shù)水平,高分辨率和寬視場(chǎng)仍然是一對(duì)矛盾體。常見的解決方法是通過圖像配準(zhǔn)和拼接的方法,也就是是利用傳感器得到多幅高分辨率的小視場(chǎng)圖像,然后對(duì)這些具有一定重疊區(qū)域的圖像進(jìn)行配準(zhǔn)拼接,得到一幅大視場(chǎng)圖像。這種矛盾在敏捷衛(wèi)星的任務(wù)中會(huì)出現(xiàn)的更加頻繁,敏捷衛(wèi)星是近年來出現(xiàn)的新型衛(wèi)星,具有優(yōu)秀的姿態(tài)加速功能,使得敏捷衛(wèi)星可以對(duì)目標(biāo)進(jìn)行更快速的瞄準(zhǔn)和更精確的掃描,大大提高了滿足復(fù)雜任務(wù)的要求。但是敏捷衛(wèi)星由于其高機(jī)動(dòng)性,會(huì)使得成像環(huán)境更加的復(fù)雜,從而影響最終的遙感圖像質(zhì)量,因此對(duì)敏捷衛(wèi)星的成像模式進(jìn)行深入細(xì)致地研究,可以定量化地分析機(jī)動(dòng)成像后的圖像質(zhì)量變化特性,而因?yàn)槊艚萏匦詭淼膱D像配準(zhǔn)拼接問題也會(huì)成為更大的挑戰(zhàn)。 通過對(duì)經(jīng)典的配準(zhǔn)和拼接算法的研究,為后續(xù)的敏捷衛(wèi)星的圖像拼接約束分析提供理論支持。圖像配準(zhǔn)拼接技術(shù)可分為兩類：基于區(qū)域的圖像配準(zhǔn)技術(shù),如序列相似性檢測(cè)算法、交叉相關(guān)相似性度量函數(shù)和傅里葉變換算法；基于特征點(diǎn)配準(zhǔn)算法包括sIFT特征點(diǎn)提取算法和Harris角點(diǎn)提取算法。對(duì)這些配準(zhǔn)算法的研究可以從圖像角度對(duì)敏捷模式帶來的影響進(jìn)行分析,同時(shí)也為基于遙感景物內(nèi)容特性的配準(zhǔn)算法提供了基礎(chǔ)。 通過構(gòu)建幾何模型、輻射模型、相機(jī)模型和大氣模型,來模擬分析敏捷衛(wèi)星的機(jī)動(dòng)成像模式,并且提出梯度信息熵、仿射退化度和立體結(jié)構(gòu)相似度等新穎評(píng)價(jià)指標(biāo)來預(yù)估圖像質(zhì)量。基于四個(gè)模型對(duì)敏捷衛(wèi)星的成像模式進(jìn)行分解,使得可以通過衛(wèi)星軌道等基本參數(shù)來定量地分析衛(wèi)星圖像的各個(gè)參數(shù),比如分辨率、幅寬,以及后續(xù)的圖像質(zhì)量指標(biāo)。在理論模型的基礎(chǔ)上,編寫了“敏捷衛(wèi)星成像仿真和質(zhì)量分析“軟件,可以對(duì)敏捷衛(wèi)星進(jìn)行各個(gè)模型的分析,同時(shí)可以對(duì)圖像退化進(jìn)行仿真模擬。 利用圖像復(fù)原和圖像融合的預(yù)處理技術(shù),改善敏捷成像下可能帶來的圖像模糊和顏色退化現(xiàn)象。針對(duì)圖像復(fù)原技術(shù),首先分析了典型的復(fù)原算法,然后提出了基于FOE模型的圖像復(fù)原算法,最后研究對(duì)比了圖像復(fù)原技術(shù)對(duì)于特征點(diǎn)提取技術(shù)的影響。針對(duì)圖像顏色退化,本文分析了典型的圖像融合算法,并對(duì)融合后的遙感圖像進(jìn)行配準(zhǔn)拼接,提高其目標(biāo)識(shí)別的能力。 通過對(duì)遙感圖像景物內(nèi)容特性的分析,提出了幾種改善性的配準(zhǔn)和拼接算法,使得遙感圖像有更廣泛的實(shí)際應(yīng)用能力。首先提出了一種基于梯度信息權(quán)重優(yōu)化的配準(zhǔn)技術(shù),利用圖像的梯度信息對(duì)圖像的特征點(diǎn)進(jìn)行權(quán)重劃分,利用權(quán)重值對(duì)特征點(diǎn)進(jìn)行區(qū)別優(yōu)化,最后將聯(lián)合的圖像拼接技術(shù)應(yīng)用到重疊區(qū)域,實(shí)現(xiàn)對(duì)敏感區(qū)域進(jìn)行高精度匹配的目的。針對(duì)遙感圖像含有豐富的內(nèi)容特性,設(shè)計(jì)了一種雙特征點(diǎn)配準(zhǔn)算法,利用SIFT和Harris短發(fā)提取特征點(diǎn)種類的區(qū)別,分別對(duì)角點(diǎn)區(qū)域密集的地方實(shí)現(xiàn)Harris特征點(diǎn)提取,對(duì)于相對(duì)平坦的區(qū)域(比如草地和水面)進(jìn)行SIFT特征點(diǎn)提取,最后得到可以應(yīng)付復(fù)雜環(huán)境下得到的推掃圖像。遙感圖像通常含有豐富的內(nèi)容,使得其特征點(diǎn)數(shù)量通常非常的巨大,應(yīng)對(duì)這種情況,提出了基于精煉控制點(diǎn)的配準(zhǔn)技術(shù),提取少量的精確特征點(diǎn)可以減少錯(cuò)誤匹配特征點(diǎn)對(duì)結(jié)果的干擾,同時(shí)可以降低對(duì)遙感圖像的處理難度。
[Abstract]:Remote sensing images refer to figure object images, common platforms such as aircraft and satellites, acquired from a remote platform using a photo-imaging load. Because of its wide range of coverage, the spectral band is rich and plays a great role in both military and civilian use. But as the area of human activity is becoming more and more extensive, the overall requirement of the remote sensing image is higher and higher, and it is hoped that the area covered by the remote sensing image is wide enough to carry out detailed and in-depth analysis and research on a large area, Support for follow-up human decision-making. However, limited to the current state of the art, the high resolution and wide field of view are still a pair of contradictions. The common solution method is to obtain a large-field-of-view image by using a sensor to obtain a plurality of high-resolution small-field-of-view images by using a sensor to obtain a plurality of high-resolution small-field-of-view images. The contradiction is more frequent in the task of the agile satellite, and the agile satellite is a new type of satellite that has emerged in recent years, has excellent attitude acceleration function, so that the agile satellite can carry out more rapid aiming and more accurate scanning of the target, and the requirements of meeting complex tasks are greatly improved. However, because of its high mobility, the agile satellite can make the imaging environment more complex, so as to influence the quality of the final remote sensing image, so that the imaging mode of the agile satellite is deeply researched, and the image quality change characteristic after the mobile imaging can be quantitatively analyzed, The image matching problem caused by the agile character can also become a bigger challenge. Through the research of the classical alignment and stitching algorithm, the paper provides the theoretical support for the analysis of the image mosaic restriction of the subsequent agile satellite. The image matching technique can be divided into two types: region-based image matching technique, such as sequence similarity detection algorithm, cross-correlation similarity measure function and Fourier transform algorithm, and feature point matching algorithm including the SIFT feature point extraction algorithm and the Harris corner point extraction algorithm. In this paper, the influence of the image angle on the agile model can be analyzed, and the basis of the alignment algorithm based on the content characteristics of the remote sensing scene is also provided. By building the geometric model, the radiation model, the camera model and the atmosphere model, the maneuvering imaging mode of the agile satellite is simulated, and the new evaluation indexes such as the gradient information entropy, the affine degradation degree and the three-dimensional structure similarity are put forward to estimate the map. The imaging mode of the agile satellite is decomposed based on the four models, so that the parameters of the satellite image can be quantitatively analyzed by the basic parameters such as satellite orbit, such as resolution, width, and subsequent image quality. On the base of the theoretical model, the 鈥淪imulation and quality analysis of agile satellite imaging鈥，

本文編號(hào)：2398187