0806514_Full MSc Project Proposal

UNIVERSITY OF GLASGOW 0806514

MSc Project Proposal 13 May 2015 Prepared for: Professor Paul Bishop and Mrs Anne Dunlop Prepared by: Peter McCready, MSc Candidate Project Timescale: May 2015 - End of August 2015 !!Title Digital building conservation: a comparative study of Terrestrial Laser Scanning and Photogrammetry. A case study of the Dougalston Estate Doocot in Milngavie, Glasgow.

! |!1 18


EXECUTIVE SUMMARY !Project Outline

The project is a digital building conservation project that will take place on the Dougalston Estate in Milngavie, Glasgow. A three-dimensional model of the interior and exterior of the B-listed Dougalston Doocot will be generated through Terrestrial Laser Scanning and will be compared to a section of Photogrammetry in order to assess the best method for digital building conservation. !Overall Objective

Digitally conserve the B-listed Dougalston Doocot in the form of a geo-referenced 3D interior and exterior model. This will be submitted to the Royal Commission on the Ancient and Historical Monuments of Scotland (RCAHMS) for their Canmore database and will be provided to local heritage and history groups for their websites. !Comparative Study

Compare two different methodologies, Terrestrial Laser Scanning and Photogrammetry, and their abilities to convey the true appearance and dimensions of the structure. This will involve comparing the differences between the two outputs, the level of observable detail, the time taken to learn how to use the appropriate hardware and software, their ease of use and the equipment required. !Project Goals

• Create a geo-referenced 3D Terrestrial Laser Scanned Model of the Doocot • Create a 3D Photogrammetric Model of a section of the Doocot • Compare the Terrestrial Laser Scanning and Photogrammetry methodologies • Offer recommendations for selecting a methodology for digital building conservation • Offer recommendations for a similar future project !Educational Goals

• Successfully perform a Terrestrial Laser Scan Traverse and capture sufficient detail • Select appropriate point cloud processing and modelling software for Terrestrial Laser Scan data • Select the correct camera for a Photogrammetry project • Successfully obtain sufficient photography for Photogrammetric analysis • Select appropriate image-processing and modelling software for Photogrammetric analysis • Perform Photogrammetric image-processing and modelling to visualise an object in 3D

! |!2 18


CONTENTS !1. Background 4 !2. General Aims 8 !3. Objectives 9 ! 3.1 Create a geo-referenced 3D Terrestrial Laser Scanned Model of the Doocot 9 3.2 Create a 3D Photogrammetric Model of a section of the Doocot 9 ! 3.3 Compare the Terrestrial Laser Scanning and Photogrammetry Methodologies 9 !4. Methodology 10 ! 4.1 Terrestrial Laser Scanning 10 4.2 Photogrammetry 12 ! 4.3 Methodology Comparison 14 !5. Project Requirements 15 !6. Work Schedule 16 !7. Report Structure 17 !8. Bibliography 18 !

! |!3 18


1. BACKGROUND !The Story of Dougalston

!The area currently known as ‘Dougalston’ located in Milngavie, East Dunbartonshire has an interesting history. In 1430 the area was granted to Lord William de Grahame, the husband of King James I’s sister, leading to the introduction of a long line of Grahams in Glasgow (Peter McGowan Associates, 2006). In 1745, a Linen Mill was set up in Milngavie and a bleach-works was established in 1763 alongside the Allander River. Both introductions were credited to the Grahams, bringing industry to the area (Peter McGowan Associates, 2006). !Lord John Glassford, a famous tobacco merchant, purchased the Dougalston Estate and began extensive development of the land through planting and building. He built the category A listed; Factor House which he used to thrill party guests, and the category B listed; Doocot that was primarily used for pigeon farming (Peter McGowan Associates, 2006). Glassford also introduced an artificial loch into the area; the ‘Dugalstone’ (Dougalston) Loch (Peter McGowan Associates, 2006). His planting led to large trees that still thrive today, as shown in figure 1.1 below right. !It is suspected that the wall located behind the tree in figure 1.1 is relatively new as it links to a significantly older boundary wall. This appears to have been where the entrance to the policies from Baldernock Road was originally located. The presence of Ha-Ha fencing that line this potential entrance, shown in figure 1.2 below, also further suggests that this was the entrance to one of Glassford’s rides that used to thrill guests. The animals on one side where the Doocot is located and the people at the other, giving the impression that danger existed as no separation was clearly visible.

! |!4 18

Figure 1.1: Tree ring dating estimates that this tree at the suspected entrance to the Dougalston policies from Baldernock Road was planted at c.1770 when John Glassford was developing the grounds.

Figure 1.2: Ha-Ha fencing that kept the natural appearance of the boundaries to potentially add to the thrill of the ride through the Dougalston policies on the route to the stables.


The Doocot is an hexagonal structure that was constructed in the late 18th to early 19th century. This ‘pavilion’ style pigeon house was constructed of red brick on the interior and covered with a more decorative stone exterior and likely hard-cast. The Doocot is topped with an open lantern that would have allowed the birds to enter and exit the Doocot but this has since been covered, possibly to limit water damage to the interior. The ledge on the exterior that lines the boundary between the roof and the top of the second level may have been part of the design to prevent rats from entering the structure. This may also have been used as a perch by birds during the day-time. !It is quite possible that the Doocot could have housed one-hundred or more birds which suggests the growth of grain on the grounds. Otherwise the birds would have disturbed a nearby field not owned by the Dougalston Estate owners, that may have resulted in tension between the families. !The Dougalston Doocot has under-gone maintenance over the years and the current state of the interior and exterior of the Doocot is shown in figures 1.3 and 1.4 below respectively.

! |!5 18

Figure 1.3: Interior - Some degradation of the brick housings and the hole to the lantern that tops the Doocot has been covered by wood, possibly to prevent water damage. The potence has also been removed.

Figure 1.4: Exterior - Water damage appears visible on the bottom of the drain pipe and possibly as a result; hard-cast appears to have been removed from covering the outer walls to help the building breathe.


Survey !A Doocot, or Dovecot as they are also known, were traditionally used to house pigeons and were often seen as status symbols that flaunted the wealth of the owner. As a result, their architecture can be unique and their decoration can be strikingly beautiful. By conserving such buildings, the practises that were used to build them and the architecture of the time period in which they were constructed will not be lost to future generations. Not only do these records of material, design and construction practices provide a means to rebuild if necessary, they are also used to connect people of the present to the past. !By digitally preserving buildings it is even easier and much faster than ever to access building records with no limit on supply and no risk of physical degradation of the records over time. Digital preservation enables the data to be stored in various forms that can be copied, edited, printed and easily distributed if necessary. This complements paper copies that are limited in number and are subject to temporal deterioration. !Two effective means of digital building conservation are the techniques known as Terrestrial Laser Scanning (TLS) and Photogrammetry. TLS uses a time-of-flight or phase-based laser scanner that is mounted on a tripod to assign XYZ values to surfaces that produce return signals. This allows a vast amount of point data, known as a point cloud, to be generated that represents the object. Photogrammetry uses a set of overlapping images of an object to generate a model by matching feature points on each image using an image-processing algorithm. The algorithm in the image-processing software monitors the movement of these points through the image-sequence, allowing it to estimate the geometry of the object (Verhoeven, 2011). !Both methods are relatively quick, can be used to obtain very accurate dimensional information for an object and can display the digital object in such a way that appears life-like. This is what makes both methods effective for digital building conservation. !However, is one method better than the other?

! |!6 18


Study Area

! |!7 18

Doocot

Figure 1.5: Geographical location of the Dougalston Estate Doocot in Milngavie


2. GENERAL AIMS !!This section defines the broad aims of the project. It should be acknowledged that the project also holds educational goals but these will not be described here - they can be found in the executive summary section on page 2.

!The overall project aims are listed below:

!• Create a geo-referenced three-dimensional interior and exterior model of the Doocot to digitally

conserve the structure. !This will be submitted to the Royal Commission on the Ancient and Historical Monuments of Scotland (RCAHMS) as a project deliverable and will be distributed to local history and heritage groups for their websites. It is acknowledged that the interior model will not display the full depths of the nesting boxes that are blocked from the line-of-sight of the Terrestrial Laser Scanner. !

• Generate two three-dimensional models of the Doocot using two opposing methodologies. !• Compare the two different methodologies and their abilities to convey the true appearance and

dimensions of the structure. This will involve comparing the time taken to learn how to use the appropriate hardware and software, their ease of use, the equipment required and the quality of output etc. !

• Provide recommendations for similar future research and methodology selection for digital building conservation. !

! |!8 18


3. OBJECTIVES !! 3.1 Create a geo-referenced 3D Terrestrial Laser Scanned Model of the Doocot 3.2 Create a 3D Photogrammetric Model of the Doocot ! 3.3 Compare the Terrestrial Laser Scanning and Photogrammetry Methodologies !The main objectives of the study allow the study to be broken down into manageable goals such as:

!• Set up a Control Network using Static GNSS that can be used to transfer real-world XYZ

coordinates to the captured data. !• Process the Control Network data using Leica GeoOffice 8.1 to apply corrections to the raw data. !• Perform a Terrestrial Laser Scan traverse that covers the interior and exterior of the Doocot using

the Leica ScanStation C10 Laser Scanner. !• Process the Terrestrial Laser Scan point cloud data to produce a three-dimensional model of the

structure using Leica Cyclone 8.1 and another modelling software such as Autodesk 3DS MAX. !• Obtain photography of the interior and exterior of the Doocot for Photogrammetric analysis using

a Digital Single-Lens Reflex (DSLR) Camera. !• Process the photography using Agisoft PhotoScan software to match the images using reference

points and create a model of sections of the interior and exterior of the Doocot. !• Compare the abilities of Terrestrial Laser Scan and Photogrammetry as a means for digital building

conservation. !!Note: as the internal structure of the Doocot is very complex the use of Photogrammetry to model the interior may be extremely time consuming and its execution may not be fully appropriate to the timescale of the project. Therefore, a small section of the structure may be used for Photogrammetric analysis in order to successfully complete the project within the set timescale.

! |!9 18


4. METHODOLOGY

This section details project considerations and the methodology for data capture, data processing and comparison of the data. !Project considerations (California Department of Transportation, 2011): • Safety (Risk Assessment) • Desirable project deliverables • Project time constraints: May 2015 - End of August 2015 • Structure size, complexity and detail requirement • Possible area disturbance (e.g. golfers, animals knocking tripods etc.) and ideal observation times • Weather forecast (TLS noise) and atmospheric conditions (GNSS error source) at planned observation time • TLS system: University-owned Leica ScanStation C10 Laser Scanner

> Availability: dependent on timetabled slot, probably second week of June 2015 > Accuracy required: (TBC) - will need higher accuracy for important sections (e.g decorative stone-work) !

A Control Network will be set up using Static GNSS that can be used to transfer real-world XYZ coordinates to the captured data. This could be checked for blunders using RTK GNSS. ! 4.1 Terrestrial Laser Scanning The following steps outline the process for performing the Terrestrial Laser Scan survey:

1. Setup the control network using Static GNSS on the Reference Object (RO), Station A and Station E. Note: Station E will provide a coordinate check half-way around the traverse. 2. Set up and sight to HDS 6” targets positioned on the back-sight and fore- sight stations with the Terrestrial Laser Scanner to position scanning station. 3. Set scan job parameters - horizontal field of view, vertical viewing angle and scanning resolution at set distance. 4. Perform scan test to calculate potential scan time. 5. Run scan. 6. Move to next station. 7. Return to step 2. until the exterior and interior scans are complete. 8. Clean and model the point cloud data using TLS processing software ! Note: More than one scan of the interior of the Doocot may be required as the scanner does not have full vertical view. Two or three interior scans may be required to obtain decent coverage. Also, detail of the back of the doocots that are above or below the scanner line-of sight may not be fully captured in the point cloud. Modelling will interpolate such areas that lack sufficient detail meaning complete coverage is not required.

! |!10 18


Figure 4.1 below shows the proposed exterior traverse stations and the RO location.

!!!!!!!!!!!!

Figure 4.2, right, shows the proposed interior traverse stations. This may be changed to two traverse stations as a third station may not be needed. Stations G, H, and I link to the exterior traverse through the position of Station C. This will easier allow scans from both traverses to appear in the same ScanWorld for modelling. Higher scanning resolution may be used on some areas of interest to capture finer detail (e.g. the decorative stone work on the Doocot exterior). ! Note: the number of stations outlined above is subject to change as the project plan progresses.

! |!11 18

Figure 4.1: Proposed Exterior Traverse Stations

Figure 4.2: Proposed Interior Traverse Stations


4.2 Photogrammetry

Photogrammetry involves the creation of a three-dimensional model pieced together from photography. 3D visualisation in such a way can involve extensive data capture and long data processing times. ! This project will use the Structure for Motion (SFM) technique to obtain photography of the structure and generate the final 3D visualisation. This technique can involve the target remaining static and the camera moving relative to it, taking images as it transits the scene, or the object is placed on a rotating platform with the camera remaining static. For this project, the camera will move relative to the static building, taking images around the Doocot that will be ‘stitched together’ using the Agisoft PhotoScan image processing software. The SFM process that is used to create the 3D model of the structure is outlined below: ! 1. Firstly, the photographs are aligned using the SFM technique. - The SFM algorithm identifies image feature points (object edges or other points of detail) and monitors the movement of these points through the image-set. 2. Secondly, the information from stage one is used as input to estimate a three- dimensional (XYZ) point cloud of the target object 3. Thirdly, a mesh is created that calculates the structural geometry 4. Finally, the model is textured to appear life-like (Verhoeven, 2011) ! The following steps outline the process of performing the Photogrammetric survey: ! 1. With the use of a camera, obtain extensive photography of the structure. 2. Input a manageable image-set into the photogrammetric image-processing software for image alignment using SFM. 3. Using the image-processing software, perform image matching, feature- point-detection, and feature monitoring. 4. Use the information from stage 3 as the input for the software to estimate and render a three-dimensional point cloud dataset for the structure. ! It is anticipated that for one face of the Doocot between fifty and one-hundred photographs will be required to produce a model. This ensures good coverage of the target area, enables the best photographs to be chosen for the modelling process and also provides contingency if some photographs are rejected by the processing software. ! Commonly in Photogrammetry a camera calibration is required. However, Agisoft PhotoScan software estimates camera calibration automatically using the EXIF file that accompanies the image. Calibration parameters can also be input manually into the software if necessary.

! |!12 18


Considerations: Well focussed, evenly illuminated images will be essential in order to aid the PhotoScan software in image matching and generating the final Photogrammetric model. ! Verhoeven, G. (2011) cautions the use of inappropriate hardware for image data processing: ! “As it is obvious that high-quality reconstructions with large image files are very

resource intensive, it is advisable to work with a multicore processor, a decent amount of RAM (minimum 6-8GB) and a 64-bit operating system.” (Verhoeven, 2011, p.70) !

Parameters that should be set and remain the same throughout the image-acquisition stage include; the camera zoom, aperture, light and distance to the target. ! Placement of an object of known length (e.g. a ruler) within the photographs will aid in georeferencing the object.

! |!13 18


4.3 Methodology Comparison

This section looks to outline the comparison of the Terrestrial Laser Scan and Photogrammetry methodologies. The method used to compare the methodologies is at an early stage and will likely be added to as the project progresses. ! The following list suggests the preliminary criteria that the different methodologies will be compared on: 1. Methodology (equipment and software requirement, expense etc) 2. Point Cloud Accuracy - Compare to known measurements (tape measurements taken between easily identifiable points on the doocot). The table below gives an example of the comparison (note: not real measurements). !!!!!!!!!! 3. Data Capture Time 4. Data Processing Time ! It is expected that the criteria will be amended as further reading takes place during the last two weeks in May 2015. ! Note: placing targets on the Doocot and using a Total Station to add known points on the Doocot for comparison may be useful to assess the accuracy of the two different methodologies.

! |!14 18

OBJECT SIZE OBJECTMEASUREMENTS (m)

TAPE(TRUE VALUE)

TLS DIFFERENCE FROM TAPE

PHOTOGRAMMETRY DIFFERENCE FROM TAPE

SMALL BRICK 0.3 0.25 0.05 0.32 -0.02

MEDIUM LEDGE 0.75 0.76 -0.01 0.75 0

LARGE DOOCOT FACE

1.75 1.75 0 1.80 -0.05


5. PROJECT REQUIREMENTS !This section will provide any equipment and software that are needed in order for the project to be successful. These are given in table 1.1 below:

!!!!!!!!!!!!!!It should be noted that the unforeseen unavailability of equipment may impact the work schedule set out in the next section and in turn affect the overall success of the project. !The Doocot is currently used for storage purposes by the golf course green-keepers to store bags of fertiliser. These bags must be moved in order to position the laser scanner inside the Doocot. Assistance may be required to move this during the scanning period and return it once the data collection is complete. Help from other students on the course will be sought to minimise delays on the day of fieldwork. !It should also be noted that due to lengths of trial periods for software (commonly 1 month) such as Autodesk 3DS MAX the timing of trial start should be thought out carefully to maximise productivity.

! |!15 18

PROJECT RESOURCES

EQUIPMENT OFFICE HARDWARE PROCESSING SOFTWARE

LEICA 1200 STATIC GNSS COMPUTER LEICA GEOOFFICE 8.1

LEICA SCANSTATION C10 LASER SCANNER

EXTERNAL HARD DRIVE FOR ADDITIONAL DATA STORAGE LEICA CYCLONE 8.1

LEICA HDS 6” TARGETS x2 USB DRIVE FOR ADDITIONAL STORAGE AGISOFT PHOTOSCAN

DSLR CAMERA CLOUD COMPARE (Open Source)

TRIPOD x3 AUTODESK 3DS MAX (Trial Version)

3m MEASURING TAPE(To Take Instrument Heights)

WOODEN STATION PEGS x7

WAX PEN(To Mark Interior Station)

Table 1.1: Required Project Resources


6. WORK SCHEDULE !This section outlines the proposed work schedule for the project between May 2015 - August 2015.

May 2015 will be used for planning and proposing the project as well as researching the topic and learning to use software and equipment. For example, running trials in Agisoft PhotoScan.

Note: considerable time is anticipated to be spent on TLS cleaning and modelling due to complexity.

Work Schedule

! |!16 18


7. REPORT STRUCTURE !This section aims to give a brief outline of what can be expected to be contained within the final project report that will comprise of six chapters. Some detail of what will be covered in each chapter is described below.

!Chapter One: provides detail on the background, aims, objectives, methodology and requirements of the project. This will largely be produced from the details contained within this proposal document. !Chapter Two: evaluates the dependencies of the study such as hardware, software, building access, equipment and additional personnel requirements. For example, learning how to perform the Photogrammetry data capture method is vital for the success of the project. !Chapter Three: gives a background to the methods being used in the project, Terrestrial Laser Scanning and Photogrammetry in this case. It will contemplate the various processes involved in each method and outline any inherent problems, offering mitigative action that may be required. Detailed procedures for each method will be included within this chapter and may be documented in the appendix, if extremely detailed. !Chapter Four: will form the main body of the report and will contain details on the practical work, analysis and results. This will give an overview of the methodology while explaining any problems encountered in the field and describing any solutions that were adopted. It will then present and analyse the collected data while comparing both methodologies as a means for digital building conservation. The final results and recommendations will then be presented. !Chapter Five: is the concluding chapter of the dissertation. It discusses the project deliverables in terms of their ability to meet requirements. This section will also offer recommendations for similar future projects and closes with a few final remarks. !Bibliography: will form the final pages of the report and will cover all relevant materials utilised throughout the project. This may also contain a references section followed by an appendix. !

! |!17 18


8. BIBLIOGRAPHY !!GREAT BRITAIN. Peter McGowan Associates (2006) Survey of Historic Gardens and Designed Landscapes in East Dunbartonshire, East Dunbartonshire: East Dunbartonshire Council !Skarlatos, D., and S. Kiparissi (2012) Comparison of laser scanning, photogrammetry and SfM-MVS pipeline applied in structures and artificial surfaces. ISPRS Annals [Online] ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences 1 (3) p.299-304. Available from: http://www.photogrammetric-vision.com/uploads/8/7/4/5/8745932/isprsannals-i-3-299-2012.pdf [Accessed: 11th May 2015] !UNITED STATES OF AMERICA. CALIFORNIA DEPARTMENT OF TRANSPORTATION. (2011) Caltrans Surveys Manual. State of California: Office of Land Surveys !Verhoeven, G. (2011) Taking Computer Vision Aloft - Archaeological Three-Dimensional Reconstructions from Aerial Photographs with PhotoScan, Wiley Online Library [Online] Archaeological Prospection 18 (1) p.67-73. Available from: http://onlinelibrary.wiley.com/doi/10.1002/arp.399/full [Accessed: 7th May 2015] !!The bibliography within the final report will utilise Harvard referencing conventions and will be split into a bibliography section and a references section. The bibliography will contain all relevant material that has been used to make the final report. The references section will display only cited materials.

!~:~

! |!18 18

http://www.photogrammetric-vision.com/uploads/8/7/4/5/8745932/isprsannals-i-3-299-2012.pdf

http://onlinelibrary.wiley.com/doi/10.1002/arp.399/full

0806514_Full MSc Project Proposal

Documents

Transcript of 0806514_Full MSc Project Proposal