多光譜遙測影像自動偵測城市道路

AUTOMATIC URBAN ROAD DETECTION FROM MULTISPECTRAL REMOTE SENSING IMAGES

彭思齊(Szu-Chi Peng)

指導教授 ： 任玄

國立中央大學/太空及遙測研究中心/遙測科技碩士學位學程/碩士(2015年)

摘要

道路網路是城市化的一個重要指標，道路的信息提供了生活中的各種應用，如城市設計、導航及城市測繪，隨著都市的發展，興建了許多新道路使交通更便利，但手動更新道路資訊，是一件非常耗費人力及時間的事情。另外在颱風豪雨或強烈地震後，山區道路也經常坍方中斷，造成偏遠鄉鎮運補的困難。為了道路資訊自動更新及救災運補管理，相對地從光學遙感影像中自動偵測道路網路是一件既經濟又有效率的方式。
在過去的數十年來，大量的研究都集中於中低分辯率的影像下做道路的提取，隨著遙感影像分辨率的提高，道路特徵的細節化和複雜化，使高分辨率影像中偵測道路的方法與中低分辯率下有很大的差別，因此研究從高分辨率影像中偵測道路具有重要的理論及實際意義。
在這份研究中，主要是利用一些道路的特徵對高分辨率遙感影像進行道路網路的提取，以下是本研究的主要流程，首先，利用多尺度的Retinex算法來增強影像，再利用影像的分類取得道路的初步輪廓，接著根據道路的均質特性，光源不變性理論和多權重的方法來改善分類的結果，為了消除路面上的雜訊，利用形態學的方法來處理，最後利用形狀指標去除非道路的區域

關鍵字

遙測影像 ； 道路偵測 ； 數學形態學

並列摘要

Road network is one important index for urbanization. The information of roads provides various applications in daily life, such as urban design, navigation, and urban mapping. With the development of the city, transportation becomes more convenient and road networks also change frequently. Also after typhoons, heavy rains or earthquakes, there are usually some debris flows which may block the roads in mountainous area. Because the update of this information manually is tedious and time-consuming, for the purpose of road network updates and transportation management after disasters, automatic road extraction from optical remotely sensed images becomes an economic and efficient approach to obtain and update road networks.
In the past few decades, many approaches are proposed to extract road from remote sensing imagery, but most of studies have applied on the road extraction from low-resolution imagery. Because of the complexity of road characteristics, road extraction in high resolution images is quite different from in low resolution images. Therefore, the study of the urban road network extraction has important theoretical and practical significance.
In this study the urban road extraction of the high resolution remote sensing images based on the several basic characteristics of roads. Our proposed method includes the following steps. First, the multi-scale retinex method is used to enhance the image, and then the k-means algorithm is used to obtain the initial outline of roads. Followed by the road’s homogeneous property, the illuminant invariance theory and the multi-weighted method are used to improve the accuracy of outline. Then morphology is adopted to eliminate noise and short lines. Finally the shape index is used to remove non-road areas .

並列關鍵字

REMOTE SENSING IMAGES ； ROAD DETECTION ； MATHEMATICAL MORPHOLOGY

參考文獻

[1] D.J. Jobson, Z. Rahman, and G.A. Woodell. ”A multiscale Retinex for bridging the gap between color images and the human observation of scenes,” IEEE Transactions on Image Processing, vol. 6, no. 7, pp. 965-976, July 1997.

[7] B.A. Wandell. “The synthesis and analysis of color images,’’ IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-9, no. 1, pp. 2-13,1987.

[8] G.D. Finlayson. “Color in perspective,’’ IEEE transactions on Pattern analysis and Machine Intelligence, vol. 18, no. 10, October 1996.

[11] E.H. Land. ”Recent advances in Retinex theory,” Vision Research, vol. 26, no. 1, pp. 7-21, 1986.

[12] D.J. Jobson, Z. Rahman, G.A. Woodell, ”Properties and Performance of a Central/Surround Retinex,” IEEE Transaction on Image Processing, vol. 6, no. 3, pp. 451-462, March 1997.