摄影测量外文文献及译文

数字摄影测量和激光扫描文化遗产文档的一体化
奥尔莎娃卜克赫*·叶海亚哈拉·诺伯特
德国斯图加特大学摄影测量学院
斯图加特兄妹大街24D,D-70174
yahyashaw@ifp.uni-stuttgart.de
委员会V4工作组
关键词：文化遗产，摄影测量，激光扫描，融合，线性表面特征，半自动化
摘要：
本文在结合地面激光扫描和近景摄影测量文物古迹文件的潜力进行了讨论。

除了改进了的几何模型，一体化的目的是支持在历史场景中如边缘和缝隙的线性特性的视觉质量。

虽然激光扫描仪提供了非常丰富的表面细节，但是它并没有提供足够的数据来构造被扫描物体的所有表面特征的轮廓，即使他们在现实中都有清晰的轮廓。

在地物边缘和线性上的信息是基于图像的分析。

为此目的一个基于图像数据的集成分割过程将支持对象的几何信息从激光数据的提取。

该方法适用于基于图像的半自动化技术以填补激光扫描数据的空白，并添加新的细节，这就要求建立现场体积更现实的感知。

实验的调查与实施是基于从艾尔卡兹尼宝库获得的数据，一个在约旦佩特拉著名的纪念碑。

1、绪论
在文物古迹的文档中经常需要生成历史建筑物的三维模型。

即旅游的目的是为学生和研究人员提供教育资源。

在这些模型生成过程中，要求高几何精度，所有信息的可用性，模型中的尺寸大小和影像写实效率必须通过不用的数据收集方法来满足[埃尔-哈基姆等人，2002]。

近景摄影测量是一个在遗产文档的背景下经常采用且广为接受的技术[格鲁恩等人，2002和德贝韦茨，1996]。

在过去的十年中，这些传统的陆地方法也受益于这样一个事实:用单架相机进行数字图像采集现在是可行的。

因此，摄影测量数据采集的效率可以通过基于数字图像处
理的自动工具的集成大大改善。

此外，激光扫描已成为高质量的3D模型的文化遗产和历史建筑生成三维数据采集的标准工具[伯勒尔和马布斯，2002]。

这些系统允许快速和可靠的生成根据反射的光脉冲的运行时数以百万计的三维点。

这将使得可以在一个非常有效和密集的几何形状表面进行测量。

关于测量速率目前的局限性，准确性，或空间点的密度在不久的将来进一步的消失，因此，激光扫描似乎成为三维文件及遗产的生成演示占主导地位的方法。

尽管这些方法有重大的进展，但是还存在一定的局限性，这对最终的三维模型的质量会产生影响。

尽管目前的激光扫描仪可以快速，可靠地产生大量的点云，但这个数据的分辨率仍然是不充分，特别是边缘和线性地物是否被收集。

相反，数字摄影测量是在轮廓再现更准确的，特别是如果他们在现实中轮廓分明的情况下。

另一方面，对于表面含有不规则的和无标记的几何细节的部分图像单独建模困难的，甚至不切实际。

另外，可以手动或半自动识别被测点，但这是一个漫长而繁琐的工作，特别是如果有相当数量的点被捕获。

如果数据收集是从不同的角度实现的，才能保证像文物古迹空间复杂对象的完全覆盖。

尽管这在大多数情况下是可能的，但问题可能会导致一个事实，即建立和拆卸完成激光系统是相对耗时的。

与之相反，这种尝试用一个标准的数码相机收集附加图像几乎可以忽略。

此外,相对于激光扫描测量范围在图像集合的限制越来越少，为覆盖该对象的完整结构简化了不同视点的选择。

出于这个原因，有利于完成一个已经从激光测量的基础上产生，从该范围内的数据独立地捕捉强度图像的几何模型。

通过这些方法几何对象，基于摄影测量提供的范围内的数据由于闭塞是不可使用的。

因此，如果这两种技术在数据处理过程中被组合，那么数据采集的效率和灵活性达到最高程度将是可能的。

在我们的方法这种融合有助于提高收集三维模型的几何形状和视觉质量。

在数据收集边缘和类似裂纹线性地物信息是基于从激光数据提供的对象的几何信息图像的分析。

此外，不可访问的激光数据扫描区域原因是由于被遮盖，这就需要增加图像的半自动化评价。

通过这些方法，现场一个完整的三维特征可以生成足够清晰的细节。

在本文所提出的方法是在一个项目针对三维虚拟模型的艾尔卡兹尼宝库生成证明框架，
就是在约旦佩特拉一座著名的纪念碑。

第二节对采集和相关图像预处理和激光雷达数据进行了讨论。

这个预处理的要求主要是为了配准激光和图像数据以供进一步处理。

第三部分提出了我国模范特征提取方法用于艾尔卡兹尼宝库左门的混合动力系统。

2、数据采集和预处理
在与约旦哈希姆大学合作进行的数据的收集，它已经用于我们的学术研究。

其中一个项目的目标是在约旦佩特拉城艾尔卡兹尼宝库纪念碑生成的三维文件，这是在如图1所示。

图1. 佩特尼艾尔卡兹尼宝库正面
2.1、艾尔卡兹尼纪念碑
古代纳巴泰佩特拉城经常被称为古代的第八大奇迹。

从公元前400年到公元106佩特拉城是约旦西南部繁荣的纳巴泰帝国首都。

佩特拉的庙宇，陵墓，剧院等建筑分布在400平方英里，这些建筑被雕刻成玫瑰色的砂岩峭壁。

之后一个旅客进入佩特拉，从山上一两公里令人印象深刻的裂缝进入佩特拉，第一门面被看作是艾尔卡兹尼宝库，这被认为是在佩特拉城最有名的古迹。

艾尔卡兹尼宝库正面是高40米,保存相当完好,可能是因为它的密闭空间保护它
免受侵蚀的影响。

当阿拉伯人叫它艾尔卡兹尼宝库的时候，意味着路是过的骆驼商队的金库或税务局，而另一些人则认为，艾尔卡兹尼宝库纪念碑是一座坟墓。

艾尔卡兹尼宝库令人印象深刻的外观背后，宽大的方形房间是用岩石雕刻出来的[赛德采克，2000]。

2.2、传感器应用
对于点的采集，MENSI公司公司制造的三维激光扫描系统的gs100，法国已经在应用。

扫描仪的特点是在水平方向360度视角，在垂直方向60度视角，使能够实现全景采集。

测距是原理是由飞行测量基于绿色激光器532 nm处的时间来实现。

该系统的扫描范围允许测量距离是2至100米。

扫描仪50米的距离的光斑大小为3毫米;距离单次测量的标准偏差为6毫米，该系统每秒能够测量5000个点。

在数据采集768X576像素分辨率的校准视频快照是另外拍摄的，它会自动映射到相应的测量点。

除了激光数据，数字图像捕获也使用富士S1相机的摄影测量处理，由于它提供了一个1536x2034像素分辨率和20毫米焦距。

2.3、测量配置
因为在艾尔卡兹尼正面基于从单个站收集的数据不可能有一个完整的三维覆盖率为，所以采用三个不同的视点与5次扫描来解决单站的遮挡。

选择的观测点位置的问题作为调查这样一个纪念碑的一个重要阶段，因为潜在的传感器站是被艾尔卡兹尼周围多山的环境限制。

三个被选中的观测点，一个在纪念碑的入口区，一个在纪念碑的左边，最后一个扫描是从高出的观测点收集。

由于认为从这些位置的激光扫描仪从一个扫描点的垂直领域不可能涵盖所有的门面，左侧和顶部扫描使用来自同一个位置2个扫描点，考虑到拥有足够的重叠区域，以便后续做整合。

总而言之，五次扫描收集近500万个点。

所有获得的三维模型都使用Innovmetric公司的PolyWorks软件来处理。

在一个独立的坐标系统合并五扫描到绝对坐标系统得出艾尔卡兹尼外观模型。

使用点对应的扫描登记后，软件构建了一个非冗余的表面表示，在被测物体的每一部分只是描述了一次。

五个激光扫描的组合的结果如图2，所生成的模型在平均每2厘米分辨率有超过1000万个三角形。

图2. 5个扫描所创建的艾尔卡兹尼的三维模型
在额外的外部调查，一个360度的扫描仪在艾尔卡兹尼宝库内一个站点扫描了1900万个
点。

图3显示了此次扫描的点云与覆盖颜色信息。

图3. 360度扫描的艾尔卡兹尼的内部
3、综合数据处理
尽管由激光扫描器产生的三维模型包含了大量描述表面的三角形，但仍然难以识别和定位的表面特征的轮廓。

这种类型特征的一个图像清晰可见的例子，如图4所示。

这个数据是从艾尔卡兹尼宝库左侧门收集的。

它可以从相应的三维视网状模型如图5所示看出，这些裂缝和边缘轮廓是失去了在现有的激光数据的分辨率。

图4.艾尔卡兹门左侧
为了支持这样的细节的视觉质量，一种从激光扫描仪和数字图像数据组合的混合方法被开发。

整合所有的数据集都是配准在第一处理步骤。

这是一种提取边缘数据源对准来实现的一种算法，这种算法是由[科林乃克和弗里奇，2003]提出来的。

传感器计算出后视定向的参数后，从点云提供在现有的图像缺失的第三维来生成距离图像。

最后，基于图像数据的集成的分割过程被用于支持提取的细节和表面特征轮廓的距离图像，加上，该方法适用于半自动
特征提取的图像来缩小与激光扫描数据的差距。

通过这些方式可以添加必要的细节产生更现
实场景的感知。

在下面的段落中混合的方法将更详细地讨论。

在艾尔卡兹尼宝库的左门在图4和图5中作为示范进行处理。

图5.网状模型左门
3.1、数据配准
在注册激光扫描仪和图像数据过程的质量,追求综合处理是一个关键因素。

这个注册对应坐标可是否以在这两个系统实现。

由于对应点的准确的检测和测量是很困难的，特别是对从激光扫描点云，直线在图像和激光数据之间对应元素之间测量。

这些线是按[科林乃克和弗里奇，2003]一个形状匹配后跟一个修改的空间后方交会。

该算法将三维直线从激光数据和相应的二维图像直线提取出来，这是由两个已知点，成为一个参数化的表示。

然后通过空间后方交会确定图像的未知的外部参数。

为了解决空间后方交会问题，利用高斯马尔可夫模型的外方位参数的初始值的最小二乘算法的实现。

他的从激光扫描仪数据的直线边缘提取被简化，如果必要的线是由两个相交平面的表面定义，如它在图6中显示所示。

数字图像中的相应的边缘提取可以交互地或半自动的基于边缘的分割。

至少三个相应的直线都需要获得一个独特
的空间后方交会解决方案的。

3.2、距离图像的生成
后视坐标和取向参数被记录在激光扫描仪坐标系中，共线条件方程用来生成一个基于现有的点云的距离图像在我们的示例场景中，点云是从1厘米平均分辨率邻近视点收集。

与相应的图像（1536x2304像素）的像素生成数目相同的距离图像。

图6.两个相交的平面表面，其中之一是由网格线用于演示目的而表示
图7。

投影在对应的图像的左门的距离图像。

艾尔卡兹尼宝库门左边,三个距离图像生成的三张图片是从不同的相机生成的。

图7描述
了其中一个距离图像投影相应的影象。

图8。

提取艾尔卡兹尼宝库左侧门的三维特征，基于图像测量柱的内缘。

3.3、图像分割
分割过程用于从三个不同的数字图像自动提取线性特征的二维坐标。

用于此目的的基础上，兰瑟滤波器的边缘检测已经应用在3成像的图像。

分段轮廓的第三个维度是从距离图像提供。

图8中，出示了提取的艾尔卡兹尼宝库的左侧门的最终三维特征。

可以看出，该数据包含了所有的边缘和线性表面特征一个清晰轮廓。

总共基于特征的代表性包含143600个点，而原始点云的相同部分有11个万点。

3.4、基于图像的测量阻挡的特点
从显示的三维功能可以看到在图5中,由于激光扫描仪的位置,内边缘的门的右列没有数据。

该遮挡边缘可以基于数字图像的半自动评估被添加到对现场一个完整的数据集。

在我们的方法中，初始点的三维坐标是手动提取。

然后一个自动立体匹配已用于紧密间隔的图像，
在边缘上添加更多的点。

对于边缘的分割部分使用极线约束进行匹配。

该遮挡边缘被添加到
苏州科技学院本科生毕业设计（论文）
三维特征，如在图8的右侧部分的显示。

4、总结和结论
本文中描述的工作是作为一个正在进行的项目的一部分，,旨在强调整合基于图像测量的必要性和激光扫描仪技术以优化几何精度和视觉的三维数据获取质量历史场景。

在我们的方法中，分割过程作为提取边缘和线状地物信息的一个中间步骤，而这些细节的三维信息是从激光扫描仪提供的数据。

由两个数据源的组合，三维的形状特征可以准确地确定，从点云的解释和啮合使用可用的图像改进模型。

最后，该方法适用于基于图像特征提取半自动化。

这些功能可以被添加到激光扫描数据，以便产生完整的场景的一个更为现实的感知。

5、致谢
特别感谢约旦哈希姆大学和佩特拉地区管理局的数据收集过程中的支持。

6、参考文献
马布斯·伯乐蒂森，A.，2002。

三维扫描仪器。

学会/协会国际研讨会上扫描的文化遗产的记录，科孚岛，希腊，pp.9-12。

戴波沃克，E,1996。

从照片渲染架构建模。

博士论文，加州大学伯克利分校。

埃尔 - 哈基姆，S.，贝尔丁，A.，皮卡德，M.，2002。

使用多种技术的古迹详细三维重建。

学会/ 协会国际研讨会扫描文化遗产录音，科孚岛，希腊，58〜64。

格伦，A.，雷蒙蒂诺，F.，张，L.，2002。

阿富汗巴米扬大佛的重建。

国际摄影测量与遥感，档案第三十四卷5部分，363-368页，科孚岛，希腊。

克莱恩科，D.，弗里奇，D.，2003。

向行人导航与定位。

在第七届东南亚调查：构造工程师协会03，香港11月3-7日。

附录Ⅱ外文原文
INTEGRATION OF DIGITAL PHOTOGRAMMETRY AND LASER
SCANNING FOR
HERITAGE DOCUMENTATION
苏州科技学院本科生毕业设计（论文）
Yahya Alshawabkeh w, Norbert Haala
Institute for Photogrammetry (ifp), University of Stuttgart, Germany Geschwister-Scholl-Strasse 24D, D-70174 Stuttgart yahyashaw@ifp. uni_stuttgart.de Commission V WG 4
KEY WORDS: Cultural Heritage, Photogrammetry, Laser Scanner, Fusion, Linear surface features, Semi-automation
ABSTRACT:
Within the paper the potential of combining terrestrial laser scanning and close range photogrammetry for the documentation of heritage sites is discussed. Besides improving the geometry of the model, the integration aims on supporting the visual quality of the linear features like edges and cracks in the historical scenes. Although the laser scanner gives very rich surface details, it does not provide sufficient data to construct outlines for all surface features of the scanned object, evene though they are clearly defined in the reality. In our approach, information on edges and linear surface features is based on the analysis of the images. For this purpose an integrated segmentation process based on image data will support the extraction of object geometry information from the laser data. The approach applies image based semi-automated techniques in order to bridge gaps in the laser scanner data and add new details, which are required to build more realistic perception of the scene volume. The investigations and implementation of the experiments are based on data from Al-Khasneh, a well-known monument in Petra, Jordan.
1. INTRODUCTION
The generation of 3D models of historical buildings is frequently required during the documentation of heritage sites
i. e. for tourism purposes or to provide education resources for students and researchers. During the generation of these models, requirements such as high geometric accuracy, availability of all details, efficiency in the model size and photo realism have to be met by the different approaches used for data collection [EL- Hakim et al 2002]. One well-accepted technique frequently applied in
the context of heritage site documentation is close range photogrammetry [Gruen et al 2002, and Debevec 1996]. In the past decade, these traditional terrestrial approaches have also benefited from the fact that digital image collection is now feasible with of-the-shelf cameras. Thus the efficiency of photogrammetric data collection could be improved considerably by the integration of semi-automatic tools based on digital image processing. Additionally, laser scanning has become a standard tool for 3D data collection for the generation of high quality 3D models of cultural heritage sites and historical buildings [Boehler and Marbs 2002]. These systems allow for the fast and reliable generation of millions of 3D points based on the run-time of reflected light pulses. This results in a very effective and dense measurement of surface geometry. Current limitations regarding the measurement rates, accuracy, or spatial point density will further disappear in the near future, thus laser scanning seems to become the dominating approach for the generation of 3D documentations and presentations of heritage sites.
Despite the considerable progress of these approaches, there are still some limitations, which have an effect on the quality of the final 3D model. Even though current laser scanners can produce large point clouds fast and reliably, the resolution of this data can still be insufficient, especially if edges and linear surface features have to be collected. In the contrary, the digital photogrammetry is more accurate in outline rendition, especially if they are clearly defined in the reality. On the other hand, image based modeling alone is difficult or even impractical for parts of surfaces, which contain irregular and unmarked geometrics details. Additionally, the identification of points to be measured, being manual or semi automatic, requires a long and tedious work, especially if a considerable number of points has to be captured.
The complete coverage of spatially complex objects like heritage sites can only be guaranteed, if data collection is realized from different viewpoints. Even though this is possible in most scenarios, problems can result from the fact that setting up and dismounting the complete laser system is
relatively time consuming. In contrast to that, the effort to collect additional images with a standard digital camera can almost be neglected. Additionally, compared to laser scanning there are fewer restrictions on the range of measurements during image collection, which simplifies the selection of different viewpoint in order to cover the complete structure of the object. For this reason, it can be advantageous to complete a geometric model, which has been generated from the laser measurement, based on intensity images captured independently from the range data. By these means object geometry, which is not available in the range data due to occlusions is provided based on photogrammetric measurements.
Thus, the highest possible degree in efficiency and flexibility of data collection will be possible, if both techniques are combined during data processing. In our approach this integration helps to improve the geometry and visual quality of the collected 3D model. During data collection the information on edges and linear surface features like cracks is based on the analysis of the images, whereas information on object geometry is provided from the laser data. Additionally, areas, which are not accessible in the laser scanner data due to occlusions are added based on semiautomatic evaluation of the imagery. By these means, a complete 3D features for the scene can be generated with sufficient and clear details.
Within the paper the presented approaches are demonstrated in the framework of a project aiming at the generation of a 3D vir-tual model of the Al-Khasneh, a well-known monument in Petra, Jordan. In section 2 the collection and pre-processing of the relevant image and LIDAR data is discussed. This pre-processing is mainly required in order to coregister laser and image data for further processing. Section 3 exemplarily pre-sents our feature extraction approach using the hybrid system for the left door of Al-Khasneh.
2 DATA COLLECTION AND PREPROCESSING
The collection of the data, which has been used for our investigations, was performed in cooperation with the Hashemite University of Jordan. One of the project goals is the generation of a
3D documentation of the Al-Khasneh monument in Petra city, Jordan, which is depicted in Figure 1.
Figure 1. Al-Khasneh facade, Petra
2.1 Al-Khasneh Monument
The ancient Nabataean city of Petra has often been called the eighth wonder of the ancient world. Petra city in southwestern Jordan prospered as the capital of the Nabataean empire from 400 B.C. to A.D. 106. Petra's temples, tombs, theaters and other buildings are scattered over 400 square miles, these architectures are carved into rose-colored sandstone cliffs. After a visitor enters Petra via Al-Siq, a two-kilometer impressive crack in the mountain, the first facade to be seen is Al-Khasneh, which is considered as the best-known monuments in Petra city. The Al-Khasneh facade is 40m high and remarkably well preserved, probably because the confined space in which it was built has protected it from the effects of erosion. The name Al- Khasneh, as the Arabs call it, means treasury or tax house for passing camel caravans, while others have proposed that the Al- Khasneh Monument was a tomb.
Behind the impressive facade of Al-Khasneh, large square rooms have been carved out of the rock [Sedlaczek, 2000].
2.2 Sensors Applied
For point collection, the 3D laser scanning system GS100, manufactured by Mensi S.A., France was applied. The scanner features a field of view of 360?in the horizontal and 60?in the vertical direction, enabling the collection of full panoramic views. The distance measurement is realized by the time of flight measurement principle based on a green laser at 532 nm. The scanning range of the system allows distance measurements between 2 and 100 meters. The scanner抯 spot size is 3 mm at a distance of 50 meters; the standard deviation of the distance measurement is 6 mm for a single shot. The system is able to measure 5000 points per second. During data collection a calibrated video snapshot of 768x576 pixel resolution is additionally captured, which is automatically mapped to the corresponding point measurements.
In addition to the laser data, digital images were captured for photogrammetric processing using
a Fuji S1 Pro camera, which provides a resolution of 1536x2034 pixel with a focal length of 20 mm.
2.3 Measurement Configuration
Because it is not possible to have a complete 3D coverage for the Al-Khasneh facade based on data collected from a single station, three different viewpoints with five scans were done to resolve the occlusions. The problem to choose the viewpoint positions represents an important phase of the survey for such a monument since potential sensor stations are restricted by the mountainous environment surrounding Al-Khasneh. Three positions were selected, from the entrance area of the monument, from the left of the monument, and one scan was collected from an elevated viewpoint. Since the vertical field of view of the laser scanner from these positions could not cover all the facade from one scan, the left and top scanning were done using 2 scans from the same position, taking into consideration sufficient overlapping regions to allow for a subsequent integration. In total, the five scans resulted in almost 5 million collected points.
All the acquired 3D models have been processed using Innovmetric Software, PolyWorks. The model
of Al-Khasneh facade resulted from merging the five scans in an independent coordinate system into an absolute coordinate system. After registration of the scans using corresponding points, the software constructs a non-redundant surface representation, where each part of the measured object is only described once. The result of the combination of the five laser scans is given in
Figure 2. 3D Model of Al-Khasneh created from 5 scans
In additional to the outer survey, a 360-degree scanning from a station in the interior of the Al-Khasneh had been taken, which resulted in 19 million points. Figure 3 shows the point cloud of this scan with colour information overlaid.
Figure 2. The produced model has an average resolution of 2 cm with more than 10 million triangles.
3 INTEGRATED DATA PROCESSING
Although the 3D model produced by laser scanner contains a large number of triangles, which are representing the surfaces, it can still difficult to recognize and localize the outlines of the surface features. An example for this type of features, which are clearly visible in an image is depicted in figure 4. This data was collected from the left door of Al-Khasneh. As it can be seen from the corresponding 3D meshed model shown in figure 5, these cracks and the edges outlines are lost beyond the resolu-tion in the available laser data.
Figure 3. 360 degree scanning for the inner part of Al-Khasneh
In order to support the visual quality of such details, a hybrid approach combining data from the laser scanner and the digital imagery was developed. For integration all data sets have to be coregistered in the first processing step. This is realized by aligning the extracted edges from both data sources using an al-gorithm developed by [Klinec and Fritsch, 2003]. After position and orientation parameters are computed for the sensor stations, distance images are generated from the point cloud in order to provide the missing third dimension in the available images. Fi-nally, an integrated segmentation process based on the image data is be used in order to support the extraction of the details and the surface features outlines from distance images. Addi-tionally, the approach applies a semi-automated feature extrac-tion from images to bridge gaps in the laser scanner data. By these means details can be added that are necessary for generat-ing more realistic perceptions of the scene volume. In the fol-lowing paragraphs the hybrid approach will be discussed in more details. The
Figure 4. The left door of Al-Khasneh
Figure 5. Meshed model for the left door 3.1 Data Coregistration
Figure 6. Two intersecting planar surfaces, one of them is
represented by grid lines for demonstrating purposes.
left door of the Al-Khasneh depicted in figure 4 and 5 is exemplarily processed.
3.1 Data Coregistration
The quality of the registration process, which aligns the laser scanner data with the imagery, is a crucial factor for the aspired combined processing. This registration can realized if correspondence coordinates are available in both systems. Since the accurate detection and measurement of point correspondences can be difficult especially for the point clouds from laser scanning, straight lines are measured between the image and the laser data as corresponding elements. These lines are then used by a shape matching followed by a modified spatial resection [Klinec and Fritsch, 2003]. The algorithm transforms the 3D straight lines extracted from the laser data and the corresponding 2D image lines, which are given by two points, into a parameterized representation. Then the unknown exterior parameters of the image are determined by spatial resection. In order to solve the spatial resection problem, the least squares algorithm-using Gauss Markov Model with initial values of the exterior orientation parameters is implemented. The extraction of straight edges from laser scanner
data is simplified, if the required line is defined by two intersecting planar surfaces as it is demonstrated in figure 6. The corresponding edges in the digital image can be extracted interactively or semi-automatically based on edge segmentation. At least three corresponding straight lines are
required to obtain a unique solution for the spatial resection. 3.2 Distance Image Generation
After the camera coordinates and orientation parameters are registered in the laser scanner coordinate system, the collinearity equations are applied to generate a distance image based on the available point clouds. For our exemplary scene, the point cloud was collected from a close viewpoint resulting in 1 cm average resolution. The distance images were generated with the same number of pixels of the corresponding images (1536x2304 pixels).
For the left door of Al-Khasneh, three distance images were generated for the three images available from different camera stations. Figure 7 depicts one of these distance images projected on
Figure 7. The distance image of the left door projected on th
Figure 8. 3D features extracted for the left door of Al-Khasneh, the inner edge of the column is added from image based
measurements.
the corresponding image.
3.3 Image Segmentation
The segmentation process is used to automatically extract the 2D coordinates of the linear features from three different digital images. For this purpose, an edge detection based on the Lanser filter has been applied on the 3 imagery images. The third dimension of the segmented outlines is provided from the distance images. Figure 8, depicts the final 3D features extracted for the left door of Al-Khasneh. It can be seen that the data contains all of the edges and linear surface features in a clearly outlines. In total the feature-based representation contains 143.6 thousand points, whereas the original point cloud of the same portion has 1.1 million points.
3.4 Occluded Features from Image Based Measurement
It can be seen from the 3D features presented in figure 5 that due to the position of the laser scanner, the inner edge of the right column of the door has no data. The occluded edge can be added
based on semiautomatic evaluation of digital imagery to have a complete data set for the scene. In our approach, the 3D coordinates of initial points were extracted manually. Then an automatic stereo matching has been applied for closely spaced images to add more points on the edge. For matching within the segmented parts of the edge epipolar constraint is used. The occluded edge is added to the 3D features as it can be shown in right part of figure 8.
4 SUMMARY AND CONCLUSION
The work described in this paper was developed as a part of an ongoing project, which aims to underline the necessity to integrate image based measurements and laser scanner techniques in order to optimize the geometric accuracy and the visual quality of 3D data capture for historical scenes. In our approach, the segmentation process is used as an intermediate step to extract information on edges and linear surface features, whereas the 3D information of theses details is provided from the laser scanner data. By the combination of both data sources, the shape of 3D features can be determined accurately, since the interpretation of point clouds and meshed models is improved using the available images. Finally, the approach applies semi-automated image based feature extraction. These features can be added to data from laser scanning in order to generate a more realistic perception of the complete scene.
5. ACKNOWLEDGEMENTS
Special thanks to the Hashemite University of Jordan and Petra Region Authority for support during the data collection.
6. REFERENCES。