【摘要】：3D數(shù)字城市是視景仿真以及數(shù)字地球的基礎(chǔ)。實現(xiàn)城市數(shù)字化的關(guān)鍵是進行快速有效的城市建模。但多數(shù)建模方法都需要很多先驗信息甚至大量人力，并且手工建模很難處理大規(guī)模分布的城市元素，，所以僅依靠手工建模不容易滿足大規(guī)模城市的高效建模。 本文采用了一種基于遙感圖像的、無成本的、靈活的自動3D數(shù)字城市建模方法，其主要內(nèi)容包括圖像分割、圖像形態(tài)學(xué)運算、建筑物輪廓提取、3D模型重建、紋理貼圖等，可有效解決手工建模的不足。 在上述基于圖像的三維重建方法中，建筑物輪廓的識別與提取決定著最后的重建效果，而圖像的分割與形態(tài)學(xué)運算直接影響著建筑物輪廓的提取。文章重點論述了基于遙感圖像的建筑物輪廓提取過程及方法，并研究了圖像分割和圖像形態(tài)學(xué)對最后結(jié)果的影響。對于遙感圖像中建筑物輪廓的提取，本文采用了區(qū)域特征與邊緣信息相結(jié)合的方法：首先對圖像采用Mean Shift多尺度分割算法進行預(yù)分割，隨后再用Otsu閾值算法對其進行后續(xù)分割，初步清除圖像中的大部非建筑物部分；最后使用形態(tài)學(xué)運算對分割結(jié)果進行優(yōu)化，有效地將與建筑屋頂顏色特征及光譜特征都相似的道路分離出去。文章最后對基于建筑物輪廓的三維重建過程進行了簡要介紹。 實驗結(jié)果表明：本文所用方法可達(dá)到令人滿意的建筑物提取效果，實用價值較高；易于理解與實現(xiàn)也是本文所用方法的優(yōu)點。
[Abstract]:The 3D digital city is the foundation of visual simulation and digital earth. The key to realize the city digitization is to carry on the fast and effective city modeling. However, most modeling methods require a lot of prior information or even a lot of manpower, and manual modeling is difficult to deal with large-scale distributed urban elements, so it is not easy to satisfy the efficient modeling of large-scale cities by manual modeling alone. In this paper, a kind of automatic 3D digital city modeling method based on remote sensing image, which is cost free and flexible, is adopted. Its main contents include image segmentation, image morphology operation, building contour extraction, 3D model reconstruction, texture mapping and so on. It can effectively solve the problem of manual modeling. In the above three dimensional reconstruction method based on image the recognition and extraction of building contour determine the final reconstruction effect and the segmentation and morphological operation of image directly affect the extraction of building contour. This paper mainly discusses the process and method of building contour extraction based on remote sensing image, and studies the effect of image segmentation and image morphology on the final results. For the extraction of building contour from remote sensing image, this paper adopts the method of combining regional features with edge information. Firstly, the Mean Shift multi-scale segmentation algorithm is used to pre-segment the image, and then the Otsu threshold algorithm is used to segment the image. Initial removal of most non-building parts of the image; Finally, the segmentation results are optimized by morphological operation, and the roads which are similar to the building roof color and spectral features are effectively separated out. At last, the process of 3D reconstruction based on building contour is briefly introduced. The experimental results show that the method used in this paper can achieve a satisfactory result of building extraction, and is of high practical value, and is also the advantage of the method used in this paper.
【學(xué)位授予單位】：西安電子科技大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：TP751

【參考文獻(xiàn)】

相關(guān)期刊論文 前2條

1 劉正軍;張繼賢;孟亞賓;梁欣廉;孫曉霞;;基于分類與形態(tài)綜合的高分辨率影像建筑物提取方法研究[J];測繪科學(xué);2007年03期

2 陶文兵 ,柳健 ,田金文;一種新型的航空圖像城區(qū)建筑物自動提取方法[J];計算機學(xué)報;2003年07期

相關(guān)博士學(xué)位論文 前2條

1 楊淼;基于圖像的城市建筑物三維自動重建參數(shù)化建模方法研究[D];中國海洋大學(xué);2009年

2 王麗英;城市環(huán)境的過程式建模技術(shù)研究[D];浙江大學(xué);2009年

本文編號：2394961