[Headnote]
Because Spoornet's historical records are often Inaccurate and incomplete, it became necesary to re-surveg its mainlines. As conventional land survey methods are more expensive and time consuming, It was decided to use a laser altimetry system, a new and Innovative technology. This article by Willem Ebersnhn. Professor in Rallway Engineering at the University of Pretoria, and P B Venter. Senior Engineer. Spooenet Infrastructure.
Spoomet allocated an international laser altimetry tender and Fugro-Inpark with its FLI-MAP, a mobile laser altimetry system, was selected to conduct the survey. At the beginning of November 1999 FLI-MAP was mobilised to South Africa where Spoornet, the South Africa railway division of Transnet, awarded Fugro-Inpark BV a major contract. This project consists of surveying 3 875 km of railway track covering all fixed assets in the right-ofway. The point density covering the assets had to be at least 10 points per m2 integrated with video images. All tools and software to process the points and video to identify assets and allocate attributes had to be provided.
The purpose is to define the accurate geographic position and attributes of all fixed assets in order to build an integrated information system to manage and maintain the fixed railway assets. The first step in establishing such a centralised maintenance management system is to identify where all assets are located. The foundation for the Spoornet infrastructure applied maintenance management (IAMM) system is a relational database with the geographic location of all fixed infrastructure as the referencing system for the database.
What is FLI-MAP?
Fast Laser Imaging and Mapping Airborne Platform (FLI-MAP) is a helicopter-based, high-accuracy airborne mapping and profiling system that can cover on average 200 km per day. The basic concept is that a helicopter flies over the corridor to be surveyed, collecting precise GPS measurements, platform attitude, laser ranges, and imagery data (figure 1).
Flying at 40-50 m altitude with a speed of 40-70 km/h, the system scans the surface area and objects directly below the helicopter at a rate of 11 000 points a second, which results in approximately 10-ZO points per M2. This high point density is required to differentiate between railway assets such as rails, kilometre posts, signals, switches, electrification wires and masts.
The FLI-MAP system integrates kinematic GPS, a reflectorless scanning laser, a solid state inertial navigation system, and digital video images into a complete remote sensing survey platform. By using advanced kinematic GPS technology, an absolute accuracy of 5-10 cm. can be achieved without compromising the environmental conditions or the necessity of permits to have access to every property. FLI-MAP is equipped with two high-resolution, broadcast-quality, digital S-VHS colour video cameras and the precise UTC time is encoded on each frame of video, providing an accurate record correlation with the laser data.
The final post-processed output, including the video from the FLI-MAP system, includes XYZ positions of the laser returns. The identification of each asset is done by recognising patterns of points with spatial relationships. Thus, the FLI-MAP system integrates the latest altimetry technology into a high-tech survey tool, which can compete with conventional survey techniques both commercially and in accuracy.
Survey operations
A normal survey flight usually consists of 5-7 base stations at known locations spaced along the flight line continuously logging GPS position during the survey flight. Survey analysis after each flight consists of checking and determining base station co-ordinates, calculating helicopter flight line and co-ordinates of surface area laser returns.
Project preparations
After signing the contract with Spoornet in Johannesburg, an agreement was made with a local helicopter firm to hire a Bell 206 Jetranger including pilot, technician and fuel support. A memorandum of understanding was signed between FugroInpark and Omega Scientific Research (Pty) Ltd, a black empowerment company in South Africa, to ensure technology transfer and training of local South African personnel. This training can be divided into the following items:
basic survey techniques introduction in GPS technology and training in GPS operations LiDAR data processing altimetry project management
* basic survey techniques
* introduction in GPS technology and training in GPS operations
* LiDAR data processing
* altimetry project management
All logistic support (hotels, transport and communications) and the employment of two armed security guards were also arranged by Omega.
At the same time, all the arrangements were made in the Netherlands to mobilise personnel and equipment to Johannesburg. Fugro's Supervisor Field Operations was send to South Africa two weeks prior to the start of the project to carry out a reconnaissance and scouting survey in collaboration with the Spoornet team. The objective of this was to locate the intended locations of the base stations and lines to be surveyed before the start of the actual survey:
To ensure that the accuracy requirements of the client are met, base stations had to be spaced at approximately 25 km so that the heli copter is never more than 10-15 km away from a base station. The reconnaissance survey ensured that the base stations were correctly spaced, taking terrain constraints and local conditions, such as hazardous surroundings, into account.
Verification of accuracy compliance
Mobilisation of the equipment, including the acceptance of the installation and flight tests by a representative of the South African Civil Aviation Authorities, took half a day. Generally time should also be allowed for clearing customs and calibrating equipment for local conditions.
To validate that accuracy requirements were met, Spoornet prepared and surveyed several concrete markers along a 100-km test section. This section had to be surveyed using the FLI-MAP system before starting the altimetry survey. The differences between the static and FLI-MAP determined positions of the targets had to comply with
Spoornet's specifications: relative accuracy per scan (50 mm) and flight (70 mm) as well as the absolute accuracy between flights (150 mm).
Table I presents the results of the comparison between the FLI-MAPdetermined positions and the co-ordinates verified by the survey division of Spoornet. The co-ordinates the in table are presented in South African Lo-system (zone 29).
The accuracy of the FLI-MAP system was accepted by Spoornet and on 4 November 1999 the actual survey of the 3 875 km started.
Daily survey operations
The data acquisition was done by two to three flight passes per day, depending on environmental conditions. Before the start of the survey, base stations were set-up on top of the 'scouted' trig-beacons (figure 2). All base stations start recording their data at a predetermined time, once all have reported their readiness to the field operation manager. During the execution of the project the accuracy of the FLI-MAP survey was monitored by flying over a base station each flight session. With the high point density of FLI-MAP, several reflections will be registered on top of the base station. Because of the accurately known position and height of the base station, the accuracy and reliability of the gathered data can be checked and verified after each flight.
As high point density is one of the requirements to be able to locate all assets, the helicopter flew at an altitude of approximately 50 rn above ground level at a speed of 60 km/h, which provided a point density of 13 to 15 per m2. This density is sufficient to clearly distinguish smaller assets such as rails. The project was completed in 25 flying and 15 transport and processing days. A total of 128 base stations were used.
Quality control
As soon as the helicopter flights were completed, the collected data from all the base stations and the helicopter were delivered to the hotel where the data processor had set up two complete processing suites. Obviously the first action of the processing team was to back up all laser files and video tapes. Once the data was backed up the information was pre-processed. Pre-processing consisted of the following control measurements to check the final quality of the surveyed data:
the overlap of joining flight lines and coverage of right-of-way
the quality of the laser points and video images
point density verification by taking a 10x10 mz section randomness in the FLI-MAP data and calculating the number of laser points
comparison of the calculated heights of the base stations with the actual published co-ordinates, which ensured that the accuracy requirements were met
As all these checks needed to be completed before the next section of the survey could start, it was not unusual for the processing team to work till very late at night and in the event of problems with the data, for the survey programme of the following day to be adjusted.
Data deliverables and ortho-rectified images
Because Spoornet personnel have knowledge of the railway infrastructure in South Africa, it was decided to do the final processing in-house. As part of the contract, Fugro-Inpark provided the client with three processing systems, including the FLI-MAP processing software, and special training on the processing.
Using the geographically referenced drawing objects created for all railway assets and the FLIP7 software, the linear distance along a base track and the perpendicular offset from this track can be calculated. This provides the dual referencing system information for each asset. The asset location and attribute information is then exported to the infrastructure database. For the delivery of ortho-rectified images a special recently added feature of the FLIP7 processing software is used. This feature captures a frame of the downward video at the desired location, and the synchronisation of the video frames with the laser data ensures the correct frame is selected. The pixels of the captured image are then corrected for position, heading and height and 'fused' with the laser data to present the image as a geo-referenced orthorectified image. This not only provides a better view for the operator on the video images and thus optimises the recognition of objects and attributes, but also enables the operator to correctly determine the position of small objects as the resolution of the video images is higher than the laser point density. Not only single images can be ortho-rectified this way, the software also offers the possibility of generating automatic seamless mosaics of rectified video images along the line of flight.
This process can be highly optimised by digitising the video images in MPEG I or 2 format and storing these digital images on hard disk, CD or DVD. The big advantage of this approach is direct and instantaneous access to every video frame without waiting for the video recorder to wind the tape to the desired location. Besides saving time it also preserves the original video tapes from damage due to excessive use.
Conclusions
The barriers of traditional techniques for corridor mapping have disappeared now that FLI-MAP can provide a method to survey long corridors by collecting remotely sensed data in a precise, reliable, cost-effective and quick way without needing to physically occupy them.
With the experience gained on the Spoornet Project from the FLI-MAP system, it is now possible to survey railway lines and take an inventory of all the infrastructure components in the right-of-way in at least half the time taken before. An additional advantage is that the survey provides an accurate visualised electronic asbuild record of the right-of-way with considerable reduction in cost and with no disruption of traffic.
The high point density provides more detailed infrastructure asset component dentification and makes the FLI-MAP survey the first important step in setting up a maintenance management system.
The extracted information can be used for asset audits (inventory), depreciation, condition management and maintenance budgeting. In addition, the data of the FLI-MAP survey lends itself well to engineering applications such as planning, design, construction and operational control of train movement.
中文翻譯
在南非測量3875 m的鐵路軌道
引言：
因為南非鐵路以前的歷史紀錄不是很完全並且不是很精確，所以重新進行測量這條路的主線是很必要的。因為普通的路線測量太耗費財力並且太浪費時間，所以決定采用壹種新創新的技術—激光測量系統。這篇文章是由皮瑞爾大學的高級工程師瓦利埃博森教授所作。
正文
這條鐵路起用了激光測高法和利用FLI地圖進行環測的方法，壹個流動激光導航系統用於這次測量。在1999年11月上旬開始在南非進行FLI地圖的測定，對南非的鐵路進行區分，政府與福格公司簽訂了壹份重要的合約。這個工程包括測量3875m長的鐵路線，這幾乎包括了南非鐵路公司的所有固定資產。測量的點密度必須達到圖像處理要求的每米至少10個點，福格公司必須要提供為了識別點和錄像帶以便與識別資產並對財產的屬性進行標定所需要的工具和軟件。
這次測量的主要目的是要確定所有的鐵路固定資產的位置和屬性並且建立壹個完整的數據庫系統以便於對鐵路資產進行保護。建立的第壹步是建立壹個可以隨時調閱各部分財產位置的財產維護管理系統。這樣做的基礎是管理系統內部的應用管理系統（IAMM）是由各部分的地理位置來表示關系的數據庫參考系統組成的。
FLI—地圖是什麽？
快速激光成像和成圖飛機空降平臺（FLI-MAP）是壹個直升機基地，高精度的成圖和成像系統能在壹天之內完成200km的任務。基本的概念就是壹架飛機飛過所需要測量路線並同時利用全球定位系統測量和收集空間數據，空間角度，測量範圍和屬性數據。
飛機以40-70km/h的速度在40-50m的高度飛行，系統將以11000點每秒的比率對直升機下面的區域和物體進行掃描，大約每平方米10個點。這樣高的點密度主要是為了區分鐵路標示，如鐵軌，欄桿，裏程碑，信號燈，電線和天線等鐵路資產之間的分別。
FLI-地圖系統把全球定位系統，激光掃描系統，固定路線的慣性導航系統和數據錄像技術結合到壹個技術平臺下面來。采用高級的全球定位系統，精度可以達到5-10cm。這樣在不考慮環境和不需要接近目標的情況下就對所有的鐵路財產進行掌握的目的就可以輕易達到了。FLI-地圖技術要求的裝備有兩個高度固定點，高質量的微波傳輸系統，數碼相機，並且把精確的UTC時間附加在數據上以便使所提供的數據更加精確。
最後的處理輸出, 包括來自 FLI- 地圖系統的錄像帶,包括激光返回的 XYZ 位置。通過各個點之間的空間關系可以確認各個財產的信息。因此，FLI地圖系統把最近的測高法技術結合到高科技的測量工具中來，在精度和商業性兩方面向傳統的測量提出了挑戰。
測量操作
在進行測量期間壹次正常的飛行壹般是通過在5-7個基礎站之間不斷地沿著飛行路線通過全球定位系統進行測量。在每次通過基礎站的飛行進行測量並且決定了車站的***縱線之後，通過計算，即可得到直升飛機飛行線路和激光測量區域表面的***縱線。
工程準備
在簽約以後，從壹個地方性的直升飛機公司雇請了飛行員，技術人員和租了壹架滿油的比爾206飛機。亞米茄公司和弗儀柯公司簽訂了壹個協議。亞米茄公司負責南非地方人員的技術訓練。訓練主要分為下列訓練科目：
全球定位系統基本測量技術的訓練和全球定位系統在LIDAR測高法數據處理中的應用。
*基本測量技術
*全球定位系統技術和全球定位系統操作的介紹
* LiDAR 數據處理
* 測高法工程管理
所有的後勤人員（服務，運輸和聯系）和兩個武裝的保安人員的雇傭都由亞米茄公司來安排。
同時，所有的儀器和人事安排被確定了下來。弗儀柯在開始計劃之前，派出了技術人員去南非和當地的人員進行合作進行了為期兩周的偵察和大體調查工作。目的是為了使所經過的車站和路線進行大致的了解以便於在真正測量之前了解大致的情況。
為了保證精度，基礎站必須相隔大約25km以使直升飛機飛行不超過10-15km。壹定要確定車站位置的正確性，考慮到環境的限制和每個地方的具體情況，例如：危險的環境就要在考慮的範圍之內。
精度檢驗
儀器的檢驗，包括對裝置的驗收和通過南非航空局的飛行測試，這花費了壹天半時間。平常的時間要對儀器刻度進行改正和掃清關稅的問題。
為了使精確度達到要求，公司沿著100km的區域準備並且測量了許多鋼筋混凝土標記。這個區域，這個區域必須在整個測量之前進行FLI地圖測量，在基本點和FLI地圖控制目標之間的差異必須被考慮到。
測量規格：相對精度每掃描50mm ,70mm,絕對精度每掃描150mm
確定出在FLI地圖確定點和掃描縱線確定點比較的結果，***縱線在已知的計算結果中是已知的。
FLI-地圖系統的精確性經過檢驗之後，真正的調查從 1999年11月4日開始了。
每日工作：
經過兩到三天進行壹次航空測量工作，這完全取決於天氣的情況。在開始測量之前，基礎站開始進行信號處理。壹旦信號穩定，所有的基礎站在壹個預定的時間開始數據的紀錄。在測量進行期間，FLI地圖的精度通過在每壹個測段之間進行不斷的檢測來保證。為了保證點的密度，在基礎站附近要進行壹定的映射測量。因為基礎站的位置和高度都是已知的，這樣被測量的數據的準確度和可信度在每次測量之後就可以得到檢驗。
之所以要求點的密度較高，是為了確定所有資產的信息，直升飛機以60km/h的速度在離地面50m高的高度上飛行，這樣點密度大約就是13-15點/m2。這樣才可保證測圖的清楚程度，例如欄桿等較小的鐵路資產才可以被測量到。這次工程計劃飛行25次，並同時進行數據傳送以保證每天可以進行數據的處理。使用了128個基礎站。
質量保證
直升飛機飛行完之後，從各個基礎站采來的數據兩個飛機不同的數據被同時送到旅館以便與進行比較和監測。很明顯，處理的第壹部分主要就是所有的激光掃描文件和錄像帶。返回的信息要進行預先的處理，預先的處理包括如下的措施以監測測量數據的最後精度：
參加飛行的飛機排成壹行以保證覆蓋率；
激光和錄制圖像的精度；
計算在FLI地圖的任意部分的10*10m的面積上激光點的個數；
把***縱線的高度與基礎站有目的的進行高度比較，以保證準確性；
所有的上述工作都做完之後才能開始下壹步的測量工作，為了保證第二天的工作能夠正常進行，每天都要工作到深夜把所有的數據都處理完，並且處理所有出現的問題，這樣做很不容易。
數據的傳送和圖像的傳輸
因為南非鐵路局對所有的南非鐵路的結構都很清楚，所以有他們來做最後的數據處理工作。
按和約上所規定的，弗儀柯提供三套處理系統，包括FLI地圖處理系統，和對人員的特別訓練等。
根據地理參考書在FLI地圖軟件上畫出所有的關於鐵路財產的地物。所畫的線離鐵路的距離和相對高度來自於預先處理的數據。這樣就為鐵路資產的判定提供了雙重的參考數據。資產的位置和屬性數據被輸入到系統內部的數據庫。采用了壹個新增加的FLI地圖處理軟件進行數據圖像的修正。在捕獲壹個特定位置的圖形數據的下壹格和激光數據進行比較，然後這個被捕獲的圖像經過激光數據在距離高度方面的修正，成為壹個參考圖像。這樣就為數據的操作人員提供了較好的圖像參考，不但使得物體的屬性容易被確認，而且使操作人員對物體的位置和大小的判定提供幫助，因為圖像的象素數比激光點的密度高。軟件不但可以進行圖像的修訂，而且可以提供對由於在飛行中產生的馬賽克現象的處理，可以達到天衣無縫的程度。
這個程序可被用於MPEG1 和MPEG2 格式的圖像和在硬盤上儲存的數字圖像，如CD和DVD。它的最大優點是不需要等待它們傳輸到放映機之前就可以對錄像帶的每壹個柵格進行格式的轉換，除了節省時間以外還可以避免由於過的使用而使錄像帶遭到損壞。
結論
由於FLI地圖系統的出現，以前的關於測量路線地圖的困難現在已經消失了，FLI地圖可以通過收集影像數據測量壹個走廊形的地帶，精度更高，花費更少，數據更可靠，並且不需要直接接觸。
利用從南非鐵路使用FLI地圖系統的這個工程得到的經驗，現在測量鐵路線路並且同時得到附近地物的信息只需要花過去時間的壹半。另外壹個好處就是可以得到壹個精確的電子數據，花費少並且不阻斷交通。
在壹個維護管理系統上設定好點的密度並且進行對各種財產成分的分析使進行FLI地圖調查的第壹步。
測量得出的數據可被用於財產調查，資產折舊，資產管理和進行預算。除此之外，FLI地圖的測量數據還可用於各種工程，例如，進行設計，計劃，建造和進行火車的調度。