應用高解析度遙測資料於都市土地違規使用之監測--以台北市建物屋頂加蓋物為例

Applying Remote-Sensing Data of High Resolution to Monitoring Illegal-Landuse in Urban Areas--A Case Study of Rooftop Structures in Taipei City

張家豪(Chia-Hao Chang)

指導教授 ： 朱子豪

國立臺灣大學/理學院/地理環境資源學研究所/碩士(2006年)

https://doi.org/10.6342/NTU.2006.00422

摘要

隨著社會經濟的發展，都市人口數不斷增加，造成在寸土寸金的都市地區中，部分都市居民寧願將土地或建物作違規使用，屈居於公共安全及公共衛生條件皆令人疑慮的違章建築中。而在都市地區最常發生的土地或建物違規使用，莫過於建物屋頂加蓋物。本研究欲透過遙測資料的特性，進行高精度與高自動化的建物屋頂加蓋物偵測，以其有效減少並遏阻屋頂加蓋物的發生。
為能更有效地控制上述問題，本研究採用兩種方式：1.以高解析度遙測影像與HIS影像進行建物屋頂形狀及材質分析。藉由遙測影像能夠快速且大範圍取得地表空間資訊之特性，透過所整理的屋頂加蓋物的判釋知識，與數值地形圖中之建物邊界資訊結合，藉此判釋影像中各建物是否具有屋頂加蓋物。並針對判釋有疑異的建物，輔以影像相對高度測量與人工判釋，抽取其高度資訊並比對建物使用許可高度，藉此判斷是否有屋頂加蓋物的發生。2.利用LiDAR資料推估各建物樓高，比對建物使用許可高度判斷是否有屋頂加蓋物的發生。以建物邊界中的數值地表模型最大值減去數值地型模型的平均值，作為建物高度資訊。以此資訊與建物使用許可中的許可高度進行比對，找出樓高超出許可高度的建物，並判斷其具有屋頂加蓋物。
以本研究所提出之方法進行屋頂加蓋物判釋，影像判釋的整體判釋精度可達95%， LiDAR判釋則因地形起伏致使建物基底高度不易判斷，整體精度僅達32.5%，但若能取的各建物確切的建物基底高度，則可大幅提升判釋正確性。藉由本研究所提出之屋頂加蓋物偵測法所產生之監測成果，日後則可與建管單位之相關查報資料比對，找出違建確切發生地點、建立違建發生之時間資料庫，並可持續追蹤違建拆除成效，必能有效的節省相關單位在屋頂加蓋物查報上花費的人力及時間成本。

關鍵字

屋頂加蓋物 ； 高解析度遙測影像 ； HIS影像 ； LiDAR資料

並列摘要

Rooftop structures, the most common annexes to residential buildings in cities of Taiwan, are the lowest-cost way for dwelling units on the top floor to gain additional indoor space. However, almost all of rooftop structures are illegal for lack of construction licenses, the most common among all illegal uses of land and buildings in urban areas. As illegal rooftop structures are so many, complete spot check by local governments will incur huge amount of labor and time.
To solve the problem, this study applies high-resolution remote sensing image, HIS (hue, intensity, saturation) image, LiDAR data and features of periphery of buildings reflected on digital topographic maps in combination to accurate and highly automated detection of illegal rooftop structures. There are three criteria, one for shape, another for construction material, and the other for building’s height from LiDAR data, for interpreting a rooftop structure as illegal: the top of the structure consists of a horizontal beam and two tilting planes on the two sides of the beam (shape); the top is made of steel or alloy plates (construction material); the height information from LiDAR data is over the information from government’s records. For rooftop structures only with either of the first or second condition, auxiliary measurement and manually interpretation are used to make sure whether they are illegal.
The detecting model was tested in an area of Yangming Mountain in the north of Taipei City, resulting in overall interpreting accuracy of 95% in high-resolution remote sensing image, but the accuracy is 32.5% in LiDAR data due to the variance of terrain. It is therefore concluded that the model can be largely cost-effective in monitoring illegal rooftop structures and thus can be used to set up databases of illegal rooftop structures, including location and time, through integration with personal inspection of such structures by local governments.

並列關鍵字

illegal rooftop structure ； high-resolution remote sensing image ； HIS image ； hue, intensity, saturation image ； LiDAR data

參考文獻

內政部地政司衛星測量中心

邱式鴻 2001 從都市區立體航照影像中萃取屋頂面的實用策略，國立成功大學測量工程研究所博士論文

林文棋 2001 半自動化建物萃取之建物模型建置與操作，國立成功大學測量工程研究所碩士論文

張昆宗、何維信、史天元 1997 應用方向制約Hough轉換法於空照影像建物圖征抽取，航測及遙測學刊，2(2)，pp.15-35

陳建文 2002 彩色影像邊緣線萃取之研究，國立成功大學測量工程研究所碩士論文