Академический Документы
Профессиональный Документы
Культура Документы
Geodesy & Remote Sensing Division, Department of Civil Engineering, Indian Institute of Technology, Bombay.
Contents:
Terrestrial Laser Scanner (TLS) Problem Statement & Available Solutions Why Terrestrial Laser Scanner? Pros and Cons. Study area and data collection Methodology and work done Results obtained, Discussions and observations Future scope
Laser Scanning describes a method where a surface is sampled or scanned using laser technology.
The advantage of laser scanning is the fact that it can record huge numbers of points (millions) with high accuracy in a relatively short period of time. And for each point it will be able to provide (X,Y,Z,I) values. Terrestrial laser scanning systems are classified based on working methodology and range.
www.leica-geosystems.com/hds-C10
Although tremendous number of structures are being built around the world, they are not always being maintained well by the people.
Role of a civil engineer is not completed by just construction of a structure but further periodic maintenance is always needed. First step in any monitoring process will be Visual Inspection which is performed manually. At certain times inspector cannot survey complete area, and he may be forced to work at heights, he may loose recorded data as he may be using paper in field.
Continued
So, there are many chances for visual inspection done by human expert to be biased. This calls for adopting advanced non-contact (nondestructive) surveying techniques to obtain complete knowledge of the current state of the structures. One such revolutionary technique is Terrestrial Laser Scanner. So, in this study, effectiveness of terrestrial laser scanner as a visual inspection tool has been investigated.
Problem Statement:
In this study, applicability of TLS for determining surface deteriorations has been carried out. A study site with evidences of biological crusts has been selected and preliminary studies were performed. Presence of biological crusts mainly effects ascetic view of a structure. This can be seen usually in structures with wet surroundings (Bridges, etc.) This biological colonization develops vegetation and mosses in and around it which inturn effect the strength of a structure badly.
Study Site
Rectangular water tank (6.13m x 2.14m x 1.25m) with evidences of biological crusts located at the main administrative building of Indian Institute of Technology, Bombay is the study site considered for this study. Biological colonization may be due to climatic conditions of Mumbai or due to seepage of water from the tank. Located at the 5th floor, and path being filled by several electric cables makes it highly unsafe to have visual inspection manually.
Data Acquisition
Scan position was selected in such a way that the instrument will not get disturbed throughout the scanning process and the distance between the instrument station and the study site was around 75m. Data acquisition has been performed by using the software called Cyclone. Then, a 14 mega pixel Canon digital camera has been used in taking high resolution RGB photograph (considered to be ground truth).
Methodology
First step in processing is the cleaning of point cloud by removal of unwanted objects (portions) of the study site. An algorithm which was specially designed by Gonzalez et al., (2012) has been used on the point cloud to obtain Intensity-based orthoimages. Intensity information has been converted to pixel information. Then image classification has been performed by using ISODATA clustering method to develop a thematic map followed by accuracy assessment.
Continued
Scanning Noise removal and Orthoimage generation Classification using ISODATA
Conclusion
Continued
Defining centers of a class Pixel assigning by distance criteria Calculation of new class centers
Pixel reassigning
Results obtained
A geometrical comparison among three areas affected by biological crusts has been made using ERDAS ImagineTM software, where RGB ortho image has been considered as ground truth. Table 1 summarizes the results obtained by comparing two predefined classes biological crusts and clear concrete, from which we can observe that the results obtained are very satisfactory. Further a detailed study has also been performed by analysing three different regions of the tank, (Area-1, Area-2 and Area-3) independently which are shown in Table 2.
Classified RGB ortho-image with two predefined classes, Biological crusts and Clear concrete.
Classified Laser scanner intensity based ortho-image with two predefined classes, Biological crusts and Clear concrete.
(a)
(b)
(c)
(d)
(e)
(f)
(a) Area-1 on RGB ortho-image (b) Area-2 on RGB ortho-image (c) Area-3 on RGB orthoimage (d) Area-1 on intensity based ortho-image (e) Area-2 on intensity based orthoimage (f) Area-3 on intensity based ortho-image
RGB ortho-image
0.814 m2
2.291m2
Table 1. Geometric comparison between different classes obtained from the RGB ortho-image and the LeicaC 10 intensity based ortho-image.
RGB Ortho-image Biological crusts Area-1 (m2) Error (%) Area-2 (m2) Error (%) Area-3 (m2) Error (%)
0.089
Clear concrete
0.065
Clear concrete
0.080 18.75% 0.261 9.57% 0.160 6.97%
0.123
0.236
0.108
0.172
Table 2. Geometric comparison between three different regions (along with their sub classes) from RGB ortho-image and Leica intensity based ortho-image
Continued
During field visit to the tank, it has been observed that some parts of area-1 and area-2 were effected with coloring, and we doubt that this could be a reason for much difference in geometrical comparision of area-1 and area-2 in RGB orthoimage and laser intensity based ortho-image. So, when the contrast between the concrete and biological crusts is very low, even the laser scanning system could not measure it. The geometric comparison between the RGB and laser intensityortho-images shows better results, with errors of less than 20%.
The radiometric information provided by TLS is less frequently used due to the fact that it is greatly influenced by physical factors the atmospheric conditions (Armesto et al., 2010). Thus, we suspect that 20% of error in geometrical comparision is because of using uncalibrated intensity values. This level of accuracy is believed to be an improvement in comparison with that obtained by traditional methodologies based on qualitative data. Unfortunately, suitable data for such a comparison are not available from previously published research. The applicability of static systems can be transferred to the mobile systems.
Continued