A framework for disaster early recovery support using CIM model based on Photogrammetric Techniques Katsunori Miyamoto 1 1 Systems Engineering Department, Japan Construction Information Center Foundation, Japan Corresponding e-mail: miyamotk@jacic.or.jp Abstract In this paper, we suggest that a framework for disaster early recovery support system may give an effective solution which innovate construction production systems using CIM (Construction Information Modeling/Management) model based on Photogrammetric Techniques. When we look back on CIM promotion activity, CIM was produced as innovation of a new construction production system since 2011 by MLIT (the Japanese Ministry of Land, Infrastructure, Transport and Tourism). We aimed at IPD (Integrated Project Delivery) of BIM (Building Information Modeling) concept using IFC model data exchange through an application to CIM. MLIT and JACIC have been introducing and promoting CIM in the civil engineering fields. However, we don't reach the stage where it can be applied to general public construction because of the large-scale pilot model made by a special engineer using high price 3D CAD. In this study, we focus on contributing to preventing and mitigating-disasters for developing resilience in the lives of the citizenry, contributing to aggressive practicality in an actual site using CIM based on a collaborative tool throughout the life cycle of structure, and National fundamental infrastructure model. In order to introduce CIM in a construction site, we inspected the use of making 3D terrain model with precision easily appropriate quickly using "Photog-CAD " of the JACIC development on the disaster site with a smart phone which a field engineer usually has. In addition, we evaluated the unified model complex of photographing a surveying site, measurement data and plan data and examined the parallel use of the CIM model based on the collaborative tool of the GIS disaster database. Meanwhile we promote "Photog-CAD " for CIM in not only civil engineering fields but also the field of agriculture and push forward the application in other fields. This paper extracts useful knowledge provided by promotion activities through many field evaluations. Keywords: Construction Information Modeling, photogrammetric surveying, infrastructure lifecycle management 1. Introduction Japan is one of the countries frequently visited by typhoons and earthquakes. According to Japan Meteorological Agency, Japanese territory lies on the area where many typhoons pass and approximately 10% of shallow significant earthquakes occur. The Japanese great earthquake outbreak frequency is 293 times of the world average. Recently the local damage caused by a local concentrated heavy rain increases. As a result, social infrastructure in Japan is always under threat of damage by natural hazards. To ease local government and municipalities burden, The National Treasury bears large part of recovery cost. To receive a subsidy, applicant must prepare application materials of disaster assessment. Volume of such paper works are not necessarily proportional to amount of subsidy. It is well known that small damage occurs much more frequently than large ones. To assist preparation of application material, JACIC has been developing software named Photog-CAD. The function of this software is divided into two, composed of Photogrammetry module and Design and Cost estimation module controlled by CAD. Photogrammetry module performs close-range photogrammetry and generates terrain model (TIN) of a damaged site on the course of generating cross sections. This model will be used for acquiring DSM (Digital Service Model) of civil engineering structure. Then it is effective way to use such photos for close-range photogrammetry to obtain standard cross sections of the site. This geometric model is comprised of variable parameters. This situation of those damaged sites and cost required for restoration will be plotted on GIS for presentation and statistics purpose. Recently, digital cameras equipped with GNSS (Global Navigation Satellite System) adding QZSS (Quasi-Zenith Satellite System) are available with reasonable price, and also free GIS engines, background maps provided on line free are available. We carried out demonstration of application of such free GIS systems and free maps to show Photog-CAD derived information effectively. Then the latest model can export TIN as a dxf file. We take these models in 3D CAD and can perform 3D design and landscape design and simulation. Recently, the disasters greatly increased. In addition to this, the number of civil engineers decreased. A framework supporting efficiency and safety of requirements for a disaster recovery is regarded as demand from the viewpoint of maintenance. 1
2. Photogrammetry 2.1. Mathematical Model Most of home use compact digital cameras in market nowadays have sufficiently good quality for photogrammetry. They are equipped with optical lenses, high resolution photo-sensors. In such condition, close-range photogrammetry using home use digital camera become popular recently. Theory of estimating exterior orientation parameters is well known as collinearity condition. It requires that object, optical center of lens system and image on optical sensor must be on a straight line. By rotating coordinate axis, we can transfer to coordinates fixed to a camera. Then, we can write down mathematical model through simple geometrical consideration. (Figure 1). Figure 1. Collinearity relation The coordinate axis here is a left hand system. We apply counter-clockwise rotation first around z-axis, then around y-axis, then around x-axis to make Local coordinate axis be parallel to camera coordinate axis. Derived mathematical model is as follows (1). (1) As interior orientation model, we apply 5th order radial distortion model d k (2) 3 5 1r k2r These 2 parameters and focal distance are estimated using triplet method. Collinearity relation (1) contains 7 independent parameters. Scale and 3 independent points are located in the picture, all 7 parameters can be estimated. Thus from 4th point, when a point is located on one of the 3 pictures, Photog-CAD automatically find same point on other pictures. Internal calibration is performed with focal length, lens distortion and distortion of a photo sensor. We need at least 9 independent points to estimate parameters for internal and external orientation. Those 9 points should be distributed rather symmetrically so that calibration parameters can be estimated precisely. 2.1. Keeping Precision in a Field Practice Through simple geometrical consideration, we can estimate size of pixel image projected on the ground. When CCD is 2.33 inches in diameter, Number of pixels is 12million and aspect ratio is 4:3, size of pixel is 2.18x10-6m. Typical home use digital camera s focal length in short focus side is 4mm. Thus projected size of pixel is 5.45x10-4S m, where S is distance between object and optical center of lens system. To precisely point the center of target with 10cm diameter, empirically 15 or more pixels are to be visible. It means one scene will be smaller than 24m times 18m in this case. Precision of photogrammetry is inversely proportional to distance and parallax also is inversely proportional to distance. For example, if same camera as above is used and measurement is done in the area near the principal point, suppose base line length is 3meter, when pointing precision is 5pixels, error in z direction (direction of line of sight) in 10m distance will be 9cm. When distance is 15m under the same condition, error will be 20cm. Of course in real processing, simultaneous least square adjustment is performed and precision of estimated coordinates will be much better. This is a rough estimation to know effect of depth in z direction. From this estimation, it is strongly recommended to use some scale to control error in z direction. We recommend putting a leveling staff or rod near the center of modeled area in the picture. As is common throughout surveying practice, geometrical strength of figure should be studied. Accuracy of Photog-CAD was tested against coordinates measured with total station. Taking slope of embankment on Iruma-River Saitama, Japan as a model site, Figure 2 shows size of target area and distance between center of the site and camera. 2
2.3 Design, cost estmation When data of cross sections are exported as a file, we call 2D CAD module for designing restoration work. The 2D CAD module can refer built in database of standard work items and unit price of them for each prefecture. When we apply standard work items, we can simply select one from pull down menu and then, unit price and CAD parts are set automatically. When local work items are to be applied, we can draw CAD parts manually and register them manually. Once new items are registered, they can be used repeatedly on the PC. Figure 2. Iruma-River Test site In Table1 difference of coordinates obtained with Photog-CAD against those obtained with a Total Station (TS) are shown. Table1. Difference of coordinates (compared with TS) Focal distance 18mm 55mm Distance (m) Mean difference from TS surveyed coordinates (cm) X Y Z 42-1.7-4.2-6.4 49-2.8-0.5-0.6 42-2.4-4.9-0.5 49-0.8-3.5-3.1 56-1.9-0.8-1.6 63 0.3-5.3-2.0 70-0.7-4.1 6.0 2.2. Terrain Model, Cross Sections We can increase nodes of 3D terrain model of damaged site. Cross sections are selected and stored in storage to be used in designing restoration work. Two kinds of making of any cross sections from the orthophoto or cross sections comprised of the point of 3pieces of photographs for designation is possible. Figure 4. Design of the slope restoration work 3. Case Studies To illustrate idea of DSM acquired by Photog-CAD, we are going to show some case studies. Each case study contains DSM and typical cross section. 3.1. Case 2. Steep and vertically long slope This site is a steep slope where soil slid down after heavy rain. As is seen in the picture, slope is ca. 45degree and surface is quite smooth. Height of the slope is about 17meters above the road. This is such a site where surveyors cannot make direct survey with rods and tape. Figure 5. Steep and vertically long slope Figure 3. 3D terrain model and cross sections 3
The surface is smooth and the slope is ca. 45degree. It is impossible to climb up the slope to perform traverse survey with rods. In this case, using root of trees, corner of stones and uniquely identified leaves are used as natural targets as we cannot put targets safely on the slope. This is the case using Photog-CAD helps surveyors avoiding accident. 3.2. Case2. Taken Picture from oblique direction This site is terraced paddy field. Because each paddy is small, we cannot take enough distance In front direction. In this case, we took pictures looking up from left direction toward damaged paddy. as natural target. In cross section, we can see shape of fallen gabion. 3.4. Case 5. Cultural Property This site is the remnants of civil engineering structure. This is the masonry retaining wall of river revetment were piled flatly or in the form of a valley, and is not a damaged site. We can use Photog-CAD to acquire shape of structure of a cultural property. In this case, modeling of the structure is rather straight forward because we can find enough number of natural targets on it. We suppress the correspondence point to acquire in length and breadth mesh automatically, and modeling is possible by pressing the vertex with a mouse. Figure 6. Terraced paddy from left side direction The vertical rod at the left end and target define X-axis are located so that we take pictures from the front direction to X-axis. With this, we can estimate proper cross section. 3.3. Case 3. Taken Picture in downward direction This site is a revetment of a steep river. The base of revetment was washed and broken. The river deeply cut into ground and it is impossible to take photos from the front or even from oblique direction on nearly same height. In this case, we were forced to take photos downward from other side of the river. Figure 7. Other side of river from downward direction In this case again, we used not only targets but also corners of remained structure, root of trees, and so on Figure 8. Masonry Retaining Wall of river revetment 4. DISCUSSION With Photog-CAD, we can acquire TIN model of the surface of structure using home-use digital camera. To keep best performance, we should make models nearly rectangle shape on the picture from front direction, horizontal and vertical scale ca. 20m or less and X axis perpendicular to light axis of the camera shooting from front direction. There are some limitations associated with photogrammetry. For example objects which do not appear on the photographs cannot be modeled. When shooting angle of central direction exceeds 30degrees or so, sometimes fail to make a model. Even so, there is advantage in this technology. Digital camera recently is available in low cost, survey rods, paper targets, tapes are not costly. Because this is a kind of remote sensing, we do not have to put survey workers at risk. Compact cameras are light enough to carry in the disaster. Especially in case of disaster, reaching the site itself is a heavy task. In such context, there is an advantage for close-range photogrammetry. If we take care of problems roughly discussed in 2.1, using Photog-CAD is a simple, easy, safe way to acquire DSM with low cost. We are aiming at efficient acquisition of DSM and utilize it for practical purpose like requesting recovery cost for infra-structure. Recently we tried to 4
advance photos on line from a damaged site to our office in Tokyo and process data in our office and send result back to the office of the local government managing the structure. Although it is still experimental, in near future, this kind of usage be one of the popular ways. 5. Conclusion Photog-CAD is suitable tool to acquire DSM of structure if the object is not too large. It is simple, low cost and acceptably precise. The software is applicable not only application for restoration cost from government but also acquire digital model for reporting, managing or designing purpose. We could make 3D terrain model from photogrammetric acquisition data easily at an investigation stage and being available with the first step to the CIM realization that made use of ICT by other 3D CAD. The use of the terrain model in the field of not only the field of disaster but also other earthworks would be available easily by adding the file export function of the 3D terrain model from Photog-CAD developed by JACIC. A concept called CIIM (Civil Infrastructure Information Management) is recently proposed as a policy correction for the maintenance. CIIM is the concept that added civil infrastructure information management system and National infrastructure management system to CIM on infrastructure lifecycle management. The next Figure is the image that photogrammetric surveying data are exchanged and coordinate for CIM and CIIM, and are unified and utilized as shown in Figure 9. analysis in Photog-CAD as the reason that we expect an effect surveying method, and it calculates a lattice point of the TIN model automatically, and it rearranges the polyline of the grid for model data. In addition, the surveying accuracy is the difference with an actual value and the model data value. We could confirm precision of several centimeters. Our surveying hour was ten minutes, and analysis time was ten minutes, too. Photogrammetry is an efficient surveying method from photographs as shown in Figure 10. Figure 10. Photogrammetric Surveying by UAV On the other hand, the LP (Laser Profiler) that it has only point group data to make a structure a model is hard to use. By designating the structure edge directly as a feature point by the mouse using Photog-CAD, so it is easy to make a structure model of the river levee. In this case, the accuracy and the measurement time were like the surveying terrain results. In the field of disaster restoration at infrastructure maintenance and management phase, we explored structure of VE (Virtual Enterprise) based on infrastructure model of CIM and CIIM. We suggest the next Figure 11 based on the earlier discussions and those discussions when we image ideal CIIM's project at the maintenance phase. Figure 9. CIM,CIIM for photogrammetric surveying In Japan, recently disasters by a typhoon and the torrential rain occur frequently. Therefore, safe and effective disaster restoration measures are hurried. We acquired 3D terrain model data from 3 pieces of photographs from 3 directions effectively by the terrain data of the wide area by UAV (Unmanned Aerial Vehicle) and Photog-CAD. This is because it divide the area with the number of the wide setting that obeyed LOD (Level of Detail/ Development) automatically after the setting the coordinate axes as Figure 11. CIM-Disaster DB Platform image We acquire 3D terrain model from Photogrammetry data quickly by cooperating with the disaster spot in 5
the office. Therefore we arrange a use model with a series of processes until the collection of data, accumulation before disaster outbreak, and it is important to cooperate at the time of disaster outbreak smoothly as shown in Figure 12. other domain model. The next is the imaged plan of the JACIC managed Cloud Service Model included various services on the construction sector. We will collaborate in Web-API. As a mission of JACIC, as shown in Figure 14. Integrated Service Data bank service JACIC service Figure 12. Surveying examination of the plane model We send local surveying data to the support center on the Internet and process confirmation and analyze of the acquisition of photogrammetry data in the center by collaboration, and using cloud service and Photog-CAD, disaster restoration DB and notify field charge after having confirmed terrain model and register data with disaster situation database. We handle all of functions of the support center semiautomatically in the future in a server side and may become cloud service. Image of the efficiency support system of the disaster recovery project is shown on figure 13. 3.4. 平 常 時 での 多 目 的 用 途 の 検 討 および 紹 介 disasters occur rence emergency dispatch Paper Ledger Pole crossing surveying The acquisition rearranging of surveying data The report of disaster assessment specifications Paper Ledger JACIC CLOUD common DB Ledger DB The relative organization adjustment, observance of the management rule of relationships Using the soft as usual information disclosure Photog-CAD Figure 13. CIM-Disaster DB platform Processing analysis Documents Report disaster restoration DB JACIC CLOUD common DB We found out that this study could be applicable to Big Data CIM database based on Infrastructure Lifecycle Management. The CIM trials were just started with the innovation of the construction production system. Therefore, we would apply these results to study CIM trials. As a problem left unfinished in this study, we could not make Big Data model on Infrastructure Lifecycle information model about both common resource and domain such as 3D terrain model like LandXML standards. We should push forward the application to disasters occur rence Production of Documents Disaster assessment 3D-CAD Agreement formation CG/VR/AR Design, construction simulation Visualization tool Estimation Procurement support. digital library Web-API Data sharing Information sharing/exchange system Integrated data service CIM CIIM Maintenance DB Server Big Data Structured data + Non-structured data Electronic product control Server With Shared Product Model On Collaborative Information Systems Promotion of Construction sector ICT Systems Public works cost estimation system International Meetings Standardization Committee CIM,BIM Online Information Electronic bidding CORINS/TECRIS Construction Byproducts By-products /Surplus Construction Soil Promotion and Publication Overseas development Figure 14. JACIC Cloud image on CIM service Acknowledgments We would like to extend their thanks to those organizations who kindly allow us to carry out survey for training and test in the area in their charge. The photogrammetry module of this software is based on 3DiVision of Tokyo Denki University. Improvement of the photogrammetry module is supported by Aero Asahi Corporation under contract. The 2D CAD module is based on Jin of Civil Design and adjusted for Photog-CAD by the company under contract. Refferences [1] Kaidzu M., H.Takiguchi (2011), Simple tool for cost estimation of recovery of damaged sites Proceedings of ISARC 2011, Seoul, Korea [2] Katsunori Miyamoto (2012), A research study on the information sharing/exchange system during construction, International Conference on Computing in Civil and Building Engineering, Moscow, Russia [3] Katsunori Miyamoto (2013), Disaster recovery project support system, and assistance on site with Photog-CAD, The 15 th construction information research institute presentation lecture document, Tokyo, Japan [4] Katsunori Miyamoto (2014), A research study on for lifecycle infrastructure management with shared product model on collaborative information systems, International Conference on Computing in Civil and Building Engineering (ICCCBE), Orland, America [5]Katsunori Miyamoto (2015), CIM for sustainability appraisal of conceptual ILCM of the disaster restoration work, Proceedings of the 2 nd International Conference on Civil and Building Engineering Informatics (ICCBEI), Tokyo, Japan 6