發(fā)布時間：2018-01-10 11:02

  本文關(guān)鍵詞：無人機載框幅式高光譜影像的波段配準研究　出處：《中國農(nóng)業(yè)資源與區(qū)劃》2017年09期 　論文類型：期刊論文

  更多相關(guān)文章： 高光譜 無人機 波段配準 框幅式 匹配策略

【摘要】：[目的]高光譜成像技術(shù)在衛(wèi)星遙感平臺上由于飛行高度與技術(shù)限制等原因無法滿足當(dāng)前對于智慧農(nóng)業(yè)與精準農(nóng)業(yè)的需求,無人機平臺的出現(xiàn)可以有效地彌補這一問題,給精細化、定量化研究農(nóng)田信息提供數(shù)據(jù)支持。受平臺穩(wěn)定性,荷載能力等因素的制約,傳統(tǒng)的推掃式高光譜成像儀不適用于無人機平臺,而框幅式成像儀的應(yīng)用前景較為廣泛。但此類成像儀由于成像原理等因素的影響,獲取的波段圖像間存在姿態(tài)與位置差異,無法直接投入到后期應(yīng)用中,在投入使用前需要進行波段配準。[方法]結(jié)合傳統(tǒng)基于特征點的配準方法針對波段配準展開研究,對圖像間灰度與位置差異對配準精度的影響進行了深入分析,針對傳統(tǒng)的研究方法在匹配灰度存在非線性變換的圖像上不足等問題,設(shè)計了兩組匹配策略實驗,分別為拍攝順序的配準實驗對比與波段順序的配準實驗對比。[結(jié)果]通過實驗對比證明在引入拍攝順序的變換基準的匹配策略下可以自動且穩(wěn)定完成配準任務(wù),此方法下配準精度可以達到亞像元級。[結(jié)論]文章所提方法兼顧了波段順序與拍攝順序,匹配結(jié)果可達亞像元級,是較優(yōu)的匹配策略。
[Abstract]:[Objective] Hyperspectral imaging technology in satellite remote sensing platform because of flight altitude and technical constraints and other reasons can not meet the current demand for intelligent agriculture and precision agriculture. The emergence of UAV platform can effectively remedy this problem and provide data support for fine and quantitative research of farmland information. It is restricted by the stability of platform and load capacity. The traditional push-sweep hyperspectral imager is not suitable for the UAV platform, but the frame imager has a wide application prospect. However, this kind of imager is affected by the imaging principle and other factors. Due to the difference of attitude and position between the acquired waveband images, it can not be directly applied to the later application, and the band registration is needed before it is put into use. [Methods] based on the traditional feature-based registration method, the influence of image gray and position difference on registration accuracy is analyzed. In order to solve the problem that the traditional research method has the problem of nonlinear transformation in image matching grayscale, two groups of matching strategy experiments are designed, which are the comparison of the registration experiment of shooting sequence and the registration experiment of band sequence, respectively. [Results] the experimental results show that the registration task can be completed automatically and stably under the matching strategy with the transformed datum of shooting sequence, and the registration accuracy can reach sub-pixel level. [Conclusion] the method proposed in this paper takes into account both the band sequence and the shooting sequence, and the matching result can reach sub-pixel level, so it is a better matching strategy.
【作者單位】： 中國農(nóng)業(yè)科學(xué)院農(nóng)業(yè)資源與農(nóng)業(yè)區(qū)劃研究所/農(nóng)業(yè)部農(nóng)業(yè)信息技術(shù)重點實驗室;山東交通學(xué)院交通土建工程學(xué)院;
【基金】：國家重點研發(fā)計劃課題“玉米生長與生產(chǎn)力近地面實時監(jiān)測預(yù)測”(2016YFD0300602)
【分類號】：TP751
【正文快照】： 0引言隨著遙感技術(shù)的迅速發(fā)展,人們對地探測的精度需求越來越高,而高光譜技術(shù)具有圖譜合一、高光譜分辨率、信息量豐富等優(yōu)勢,成為了定量遙感的前沿技術(shù)之一,給定量化理解地面信息開辟了新的領(lǐng)域[|?2],其在農(nóng)業(yè)領(lǐng)域得到了廣泛應(yīng)用[23J4]。同時,由于高光譜技術(shù)不僅可以準確的，

本文編號：1404993