多航帶空載光達強度之輻射整體平差與地物分類

Abstract

空載雷射測距(光達)系統對於地形起伏的建模是具有效率且可靠的方法。光達系統根據掃描距離、入射角度及物體反射率等因素，以點雲的方式記錄物體之空間三維坐標及反射強度值。然而在不同掃描飛行航帶中，相同地物的反射強度值並不相同，探究其因在於系統與環境造成的輻射畸變，為有效應用空載光達系統的強度數據構建空間資訊，適當的強度改正與輻射正規化是重要的資料處理步驟。因此，本研究之目的為使用光達強度值及其他特徵資訊進行地物分類。 本研究的第一部份為提出一種輻射正規化的策略以縮減不同航帶間的強度差異，主要工作包含三個步驟：(1)重疊區配對，(2)連結區自動選取，及(3)補償參數計算。在平衡各航帶強度差異前，先以光達方程式進行個別航帶強度改正。平差方程式的觀測量由連結區的點雲特性自動萃取產生，其特性包含空間幾何的平坦度與輻射反應的均勻度。最後解算各航帶的補償參數並將各航帶強度完成正規化平衡。 本研究的第二部份為分析空載光達系統應用於地物分類的效益。在這部分導入多光段光達系統與物件導向式分類，研究針對四個議題進行探討：(1) 強度改正前後對地物分類的差別；(2)單波段與多波段光達系統於地物分類的良窳；(3)多波段光達之幾何特徵與光譜特徵的差異；及(4)主動式多波段光達系統與被動式多光譜影像的植生指標比較。 實驗結果顯示，本研究提出之輻射整體平差方法應用於Rigel Q680i光達系統所蒐集的九個航帶資料中，強度差異值與背向散射係數差異值均提升60%以上，在地物分類實驗中，使用Optech Titan多波段光達特徵的分類整體精度高於單波段光達系統約4-14個百分點；使用光譜特徵於草地、道路及裸地等地物類別分類，其整體精度亦較幾何特徵提升29個百分點；植生指標方面，主動式光達與被動式影像之相關係數達0.68-0.89。整體而言，多光段光達系統因較傳統單波段系統提供更多的光譜資訊，可對地物分類的精度提升挹注相當的效益。

Airborne light detection and ranging (lidar) system have appeared to be effective and reliable methods for modeling the topography of the earth surface. Lidar records the coordinates and intensity of a location based on range, incident angle, reflectivity of object, atmospheric condition, and several external factors. The intensities scanned from different flight strips are not fixed as well. Due to the systematically and environmentally induced distortions, airborne lidar intensity data requires certain correction and normalization schemes to maximize the benefits from the collected data for constructing spatial information. Therefore, in this study we utilize intensity and other lidar features for land cover classification. In the first part of this investigation, this study presents a radiometric normalization to cross over the gaps among different flight strips. The proposed scheme comprises three major parts: (1) overlapped area paring; (2) tie regions automatic selection; (3) compensated parameters calculation. The correction model from lidar equation is proceeded with each strip before balancing the intensities of all strips from a lidar acquisition. Then, the overlapped areas are pair-wise to construct the equations of radiometric block adjustment. The observations of adjustment equations are extracted automatically from tie regions in overlapped area based on the characteristics of point clouds. The characteristics include flatness of spatial geometry and homogeneity of radiometric response. Finally, compensated parameters are calculated and responded to the respective strips. In the second part, for land cover classification, the benefits of airborne lidar system were analyzed. The multi-wavelength lidar data and object-based classification are applied. This study focus on four major issues: (1) the classified differences between original and corrected intensity; (2) the evaluation of single- and multi-wavelength lidars for land cover classification; (3) the performance of spectral and geometrical features extracted from multi-wavelength lidar; and (4) the comparison of the vegetation index derived from active multi-wavelength lidar and passive multispectral images. Experimental results indicate that the delta intensity and delta backscattering coefficient of tie regions were improved up to 60% after multi-strip block adjustment using nine strips lidar data acquired by Rigel Q680i system. In land cover classification, the Optech Titan multi-wavelength lidar provided higher accuracy than single-wavelength lidar in land cover classification, with an overall accuracy improvement rate about 4–14 percentage points. The spectral features performed better compared to geometrical features for grass, road, and bare soil classes, and the overall accuracy improvement is about 29 percentage points. The results also demonstrated the vegetation indices from Worldview-3 and Optech Titan have similar characteristics, with correlations reaching 0.68 to 0.89. Overall, the multi-wavelength lidar system improves the accuracy of land cover classification because this system provides more spectral information than traditional single-wavelength lidar.