Webinar on 'Remote inspection: Part 2 - Laser scanning for remote inspection'

Remote inspection: Part 2 - Laser scanning for remote inspection 

Speakers: Paul Bryan, Gary Young and Jon Bedford 

Alice [00:04] ...using laser scanning for remote inspection. So I'll hand over to Paul, who will introduce Gary and Jon. Thank you very much for today and over to you, Paul. 

Paul [00:15] Thank you very much, Alice, and good afternoon to everybody, wherever you are. I'm up in North Yorkshire, sunny North Yorkshire today. It's very spring-like, so I hope it's similar to wherever you are in the world. Welcome to the second in the series of webinars that are focusing on geospatial technologies with a remote inspection application. 

Some of you may have been on our, sort of, original-- sorry, our first webinar, where we talked about the use of drones. We're focusing today on laser scanning, and I'll highlight that we've got another one where we're talking about what we call reality capture, which is basically the fusion of laser scanning and photogrammetry, and that's on 9th March, so I hope some of you will be able to join us for that. 

As I say, it's the next webinar in the series of Technical Tuesday Technical Conservation Webinars. You can see the link there. Obviously, you all know what the link is because you all need it to register, so I'll just go onto sort of the final introductory slide and highlight that during the presentation today, I'll be highlighting some of the guidance that we have, focusing on geospatial technologies, but if you want to know more about the guidance and research that the whole of Technical Conservation generates – there is a lot – please download the PDF from the web address on the screen. 

So without further ado, welcome to laser scanning for remote inspection. I'm Paul Bryan. I'm geospatial survey manager within Historic England. My contact details are on the slide there. I'd also highlight the three logos at the bottom. ARPAS UK are the UK Drone Association, and we're a member of ARPAS UK and we work very closely with them because, as you will have heard in our previous webinar, we are increasingly using drones within our work. The other organisation on the right is The Survey Association, or TSA, and I'd strongly recommend that if you've not been to their website, please go to it because they have a host of information on the professional application of geospatial survey with a number of sort of downloadable guides as well. 

In the middle is GEO Business 2021. They're not an organisation. It's a conference, and for us geospatialists, if there's such a term, that's the event that we tend to go to. Like many it's been cancelled due to Covid, but hopefully, they'll be gathering again at the ExCeL Centre in November, so I'll highly recommend that you look at attending that. 

This webinar will be given by myself but also with Gary Young, who's our geospatial analyst and he's on the image on the left there, and my other colleague Jon Bedford, who's our senior geospatial analyst. You can read what we'll be talking about on the screen there, but it's probably best if I sort of crack on, because we have got quite a lot to get through. 

First of all, if Alice can bring in the first poll, and while she's doing that I'll try and see what we're actually asking you. It's to gauge some idea of your use of laser scanning. Thank you very much to everyone for quickly voting, but it's to see 'How many of you already use laser scanners', 'How many of you don't already use laser scanners but commission it from contractors', 'How many don't use laser scanning at all but are learning more' – very interesting that that's ahead at the moment, and 'Others who prefer not to use laser scanning and use something else'. We've got one vote for that. So I think it's stabilised there, Alice. Shall we leave it at that? Yeah? So really interesting that it's mainly 'Don't use laser scanning but keen to learn more', so a good advert for the webinars, but we have a number of people who already use laser scanners, so I hope you agree with what Jon, Gary and myself have to say. 

Laser scanning, we have a guidance document there, which you can download, but I thought I'd start off with a couple of definitions of laser scanning because technologies like photogrammetry, sometimes they need a definition for people to understand. But laser scanning, we have two definitions within the survey we tend to refer back to. The first one of these, you can read it on the screen, but I've actually highlighted the key words. It's about generating 3D coordinates automatically in a systematic pattern at a high rate and in real time. So you get a sort of flavour of what the technology provides. 

A more recent definition provided by Pierre Grussenmeyer from University of Strasbourg highlights that it's an 'active', in other words it sends something out, in this case a laser beam, so it reacts to what comes back. It's fast, it's automatic, it uses lasers – key there. These days they are class one lasers, so they're of the eye safe type so we don't need to worry too much about the health and safety aspects of shining lasers around public spaces. But they're all about measuring without contact. They're a remote, non-contact solution. So very useful for where it's difficult to actually gain access. Generating dense regular patterns of 3D coordinates of points on surfaces, and this is where the term 'point cloud' is often referred to when you talk about laser scanning. 

I've been a surveyor for nearly 40 years now, so I've been in the industry long enough to remember when we first encountered laser scanners in the UK. My experience – this is my experience – was they first arrived in the late 1990s, and I've highlighted two scanners there. The first one, which is the one on the top, is the Cyrax 2400 laser scanner that was manufactured by Cyra Technologies over in America. They were one of the two pioneer companies in the field of what we call 'time of flight' laser scanning. I'd highlight the scanner, which is the big black box on the tripod, but also all the ancillary equipment that was needed back then to actually use the laser scanner. Compare it to what we are showing later on in terms of laser scanners that are certainly not miniaturised, but they're getting smaller and smaller. 

The one on the bottom is from the Austrian company RIEGL, and that scanner is actually the LMS-Z210, and if you look at modern RIEGL laser scanners, they don't look that dissimilar to what that one is. But once again, there's a nice black box on the trolley system there. 

In English Heritage, bearing in mind that we used to work for English Heritage up till 2015. The first use of laser scanning was within a measured survey of the Iron Bridge in 1999 to commemorate the turn of the century. And that work was undertaken by UK Robotics, based over in Salford, and it used the predecessor to the IMAGER laser scanning system from Z+F. And once again, you can see the scanner, which is the gold box on the tripod, but look at the big cable that is connecting it to its power supply and the actual desktop computer that was needed back then. So they are quite revolutionary devices, these, but they've certainly developed leaps and bounds over the past few years. 

Within our world of heritage, certainly notable projects were in 2002, where we used the Cyrax 2500 to undertake some survey work in Greenwell's Pit in Grimes Graves, and you can see a slice through the data. Even in 2002, you can imagine how interested we were in the technologies and the potential applications for them. And then of course, we have Stone Henge in 2011/12, which was sort of a significant milestone in the use of laser scanning, certainly in the UK, and it contributed an awful lot of information on that monument. 

But looking at what we call terrestrial laser scanners, because they tend to be sort of the dominant scanner technology used in heritage, typically sitting on a tripod, they all work in a similar way. They emit a laser beam. It tends to come out from one side of the scanner, is deflected by a mirror that's rotating, and it measures the distance between the scanner and the surface by one of two means. One of them is time of flight, so it's actually calculating how long the beam took to go out and come back, and then it can calculate the distance. 

The other one is what's called phase comparison, so it's actually looking at the change of the wave that comes back and working out the distance that way. If you combine that with the angle of rotation of the scanner plus the angle of rotation of the mirror as it's going around, you can generate a 3D point, bearing in mind these scanners are working extremely fast, and these days a modern laser scanner can collect a 3D scan in less than two minutes. It's doing this 3D geometric calculation many, many, many times a second. But at the end of the day, it's capturing a 3D point that generates what we call the point cloud. 

So terrestrial laser scanners in the world of heritage. Very suitable because the ranges go between half a metre, so it means we can get into small spaces and capture data but also over longer ranges as well, typically up to 300 metres, although most of our work, we tend to do up to about 100 metres. The accuracy, it varies between, say, 1mm and 10mm. Most of the time, our data is round about, sort of, 2mm or 3mm. You'll hear that from Gary and Jon later. 

These days, scanners can capture more than just 3D points. All scanners, or most scanners, can actually capture an image. Not all cameras are the same. I will say this. So some are better than others, and now we're starting to see other sensory devices, like a thermal camera, attached to the scanner so it can collect thermal data at the same time. So typically from the Z+F 5016 that you see in the bottom there. 

So for heritage they're great, for recording and allowing the remote inspection of heritage buildings and structures, but they are expensive devices. I don't think the manufacturers will mind me saying that, because typically they're above 30,000+, so they are an expensive solution. You can hire them. You can rent them. You don't have to buy them, but they are quite a substantial investment. 

Over the past few years, we've seen an increasing use of mobile scanning technologies, and the image you see on the screen there is what's called the GeoSLAM Revo system. It's a handheld scanner, and it works by having a laser, a two-dimensional laser, which effectively comes out from the bottom of the top. Sorry, that's a bit of a daft way of defining it, but if you look at the central bit of the scanner where it becomes sort of plain black leading up to a gold colour, that's where the scanner beam comes out. So as it's rotating round, it's actually capturing 360-degree data, and of course, as you're moving through a site, it's building up the data as well. So they certainly have heritage applications. 

They typically use a technology called SLAM, simultaneous localisation and mapping. I haven't got time to actually go into that in any great detail. You'll see an image on the screen there that gives you a suggestion of how this scanner and the IMU, the inertial measurement unit, is allowing real-time 3D data capture and generation. So if you need to know more, please download the document that you see on the screen there. 

And the range of these is not so far, 30 metres, possibly more with some of the later ones. The point accuracy is not the same as static scanners. It's an order of ten or more worse, 30–40mm, so it's sometimes described as slightly fuzzy data because of that. Capture rates are not quite as fast, and these days they capture in colour as well. But as shown by Lizzie, our apprentice, you can capture data as you walk through a structure, so it certainly has benefits there. You can see an image that Lizzie has captured there. 

So they're excellent at recording heritage spaces. They're not particularly cheap. They're £25,000+, but if you're after the architectural detail, the fine detail within a building or in a tunnel, you probably need to look elsewhere because they're not designed to actually capture that level of information – well, not yet. I'm hoping that the manufacturers will take note and bring out something in due course. 

We're also in the world of backpack laser scanners as well, as shown by my colleague David Andrews there, very happy to wear the backpack, obviously, on that day. And you can see there, you've got a scanner combined with GNSS, or GPS receivers – 360-degree imaging cameras. So you've got a collection of sensors that enable you to actually walk through a heritage space, and you see the range, not much better than a handheld scanner – 50 metres. The relative point accuracy is better. The capture rate is far superior, collecting much denser data, and of course, with multiple 360 cameras, you're capturing a lot of colour. 

So once again great for heritage spaces, rapid recording of heritage spaces, but they're not-- well, they're not cheap. I've said £125,000 there. Yeah, you may want to contact the likes of Leica or RIEGL just to see what they're selling them for at the moment, but they are an expensive solution, but they do enable an entire site like Lincoln Bishops' Palace as shown there, an English Heritage site, captured in one day of just simply walking around the site. 

So that's enough of me for the introduction. I'm going to hand over to my colleague Gary to actually take you through how he undertakes a survey on site, but anyway, I'll hand you over to Gary now. 

Gary [16:20] Hi, I'm Gary and as Paul said, I'm a geospatial surveyor analyst here at Historic England. As part of my role, I regularly undertake laser scanner work at the heritage sites, ranging from smaller jobs, where it could be individual artifacts, right up to larger structures, inside and out. Many factors need to be considered in order to successfully complete this type of work, and I'm going to be talking to you about how we control a laser scanning survey as well as the methodology in settings you need to consider to produce a useable final product. I'll just have a quick look at the poll answers. I'm glad no one's gone for the obvious clanger at number 3. That's a good sign. OK, I think that's evening up there. I'm glad to see most people agree with me, that it's the most important part. 

So if we move on to-- So control network establishment. That's going to be the first element of any survey. You need to consider the control network you're going to need to successfully complete the work. The control network will usually consist of a series of permanent or semi-permanent control markers, which form the basis of all the horizontal and vertical control on the side. These markers can be anything from permanent concrete markers in ground anchors through to simple nails and pegs, all depends on the requirements of the project and the nature of the site. 

For most jobs, the first step will be to establish a primary baseline on site. This is a pair of indivisible control marks, which will form the basis of any additional control which may be required. These days, Ordnance Survey use a system called OS Net, which is a highly accurate and resilient network of global navigation satellite system, or GNSS space stations, which are spread across the whole of Great Britain and they form the basis of the UK's national coordinate system. OS Net has superseded the traditional Ordnance Survey control networks of the concrete triangulation pillars and benchmarks that you might be familiar with. And those are no longer maintained or used by Ordnance Survey. 

So by using our own GNSS receivers on site, in combination with data from this OS Net, we can position our primary markers onto the National Grid, with an accuracy ranging from +/-30mm down to +/-10mm, depending on the occupation techniques we use. Thanks to improvements in the vertical accuracy of GNSS receivers, this also provides the best way of tying our control to Ordnance Datum Newlyn, or ODN, which forms the datum point for all height information on the national coordinate systems. So the days of having to level several kilometres from a physical Ordnance Survey benchmark are happily long gone. 

All measurements taken on the National Grid have to take scale factor into account, so any grid system which covers a large area, such as the National Grid, needs to take into account the curvature of the earth, and the scale factor is the adjustment you need to make to a measurement made on the ground in order for it to match the same distance shown on a National Grid plan or map. 

Now, laser scanners don't apply this scale factor to their measurements. So what we need to do is produce a transformation, allowing us to define a local grid system where the scale factor has been removed from the data, but it's still orientated to north and associated with ODN, so the height, the information is still valid. 

It's possible to create a local grid where the data's aligned to a building or any feature, should that be required, but generally in the work we do, we tend to keep the real-world alignment. The transformation can work both ways, so should we wish to transform data back onto the National Grid once we've extracted line work and things from it, then that's possible. 

So our control network allows us to position our data in the correct place, but that's not its only purpose. Arguably more important is our control network's role in the quality assurance we can provide to our clients. Without constraining the survey to fix control, we've got very few ways of understanding and capturing the errors in our observations. All survey work is subject to a risk caused by many factors. Proper control allows us to understand the level of error we've got and to ensure it's kept as low as possible. Without this information, we've got no way of proving that we've met the accuracy requirements specified for the project, and our data would not be reliable if we can't prove that. 

Another important aspect of the control is repeatability, both in terms of ensuring the data captured from the same control gives us the same measurements within the error of the instrument used, as well as allowing for future visits to be on the same control network, meaning that data from multiple site visits can be combined over time. And thanks to the fact that the primary control is referenced to the external OS Net, it's also possible to replace any markers that are destroyed, ensuring that the grid systems set up on individual sites are robust and can't be lost. 

So you've got your control established, and you can now begin the actual laser scanning. So there are several settings within the scanner that you're going to need to consider depending on your requirements. The first to consider is the required point density. If you look at the image on the top right-hand side of the slide, you'll see a screen grab of the same subject scanned at two different densities. The image on the left has a density of 50mm at 10 metres, which means if the scanner was 10 metres from the subject, the points would be 50mm apart. This scan took 20 seconds to capture a full 360-degree image from the scanner. 

The image on the right was captured at 0.8mm at 10 metres, and the 360 scan at this density took 54 minutes, so as you can see, the level of detail captured on the right is massively superior to the image on the left, which depending on the size of your monitor, could possibly be so sparse that you can't actually see it. So if I just blow up a section to give you an idea. Hopefully, you can see how spread out the points are now in that image. 

So this increasing data does come with a significant time cost, and you'll need to consider how much detail you actually need for the requirements of your job. Somewhere in the middle is usually preferred in order to capture a good level of detail without being a prohibitively long process. 

You'll also need to consider the distance you'll be from the subject. The closer you are, the lower density you can use to produce a similar effect. Conversely, if for whatever reason, you need to scan from further away, then you'll need to increase the density in order to ensure that you capture sufficient detail. By default, the scanner will capture a 360-degree image around itself, although it's possible to restrict this to a user-defined field of view, which is another way you can save time, as you can avoid scanning areas which you aren't interested in. 

Alternatively, you can use window scans, where a small area of the scan is selected, and you can repeat that at a higher density, so for important areas, such as a particular area of stonework that's particularly well carved or particularly interesting that your more interested, you can do that at a higher density.  

If you're only interested in the 3D shape of a subject, then you'll be able to capture a point cloud on its own, and you'll see the data in a similar way to the middle and right-hand images of Mount Grace Priory. At the bottom of the slide, this is known as intensity view, and it's a measure of the strength of the reflected laser beam that's represented in the colour of the point. This can be viewed as greyscale, as on the right, or a multi-hue, like in the centre. 

However, for most heritage work, we're also interested in the capturing of the real-world colour and the texture of the subject. This is achieved with either external or an integrated camera that captures a series of images which are used to apply an RGB colour value to each individual point in the point cloud. The scanning and processing of the data will take longer if any imagery is used, so again, something you need to consider if you require it or not. 

Capturing imagery introduces a whole extra set of settings, which need to be considered. Firstly, you'll need to ensure that the white balance is correctly set for the environment. Otherwise, the colour may not display correctly, and you'll also need to consider the resolution of the imagery you're capturing and make sure it's sufficient for the subject. We tend to use HDR imagery as well. Both of these factors have time implications again but can provide you with better quality data and more options when it comes to processing the data. 

So now that the scanner's set up correctly, you can start to consider how you're going to capture the subject efficiently. This means we want to ensure we have as much coverage as possibly whilst not undertaking unnecessary scans, which can mean we spend longer needed on site, as well as increasing the amount of data that we have to manage. Laser scanning jobs will generally be big enough without having extra data that doesn't improve the product included in there. 

We need to carefully consider where we'll position the scanner for each individual scan to ensure that we minimise the presence of voids in the data. These are effectively shadows in the data where something has blocked the scanner's line of sight. By undertaking multiple overlapping scans, we can build up a complete picture of the subject and eliminate the voids, which are inevitably going to be present in the individual scans. 

As I mentioned on the previous slide, the distance you are from the subject will have a direct bearing on the density of the data collected. Therefore, you need to consider this when you're positioning the scanner. You'll also have to consider that the accuracy of the individual points is also directly affected by the distance they travel. For example, the scanner pictured here, which is like a P40, has a 3D positional accuracy of +/-3mm at 50 metres, but this drops to +/-6mm at 100 metres. It follows that the closer to the subject you are, the more accurate the data collected is going to be. It's possible to have the scanner automatically fill their outpoints past a certain range, as well, to ensure that all of your points meet a certain accuracy. 

So we'll move on to registration. So having completed your site work, you now need to return to the office and start processing the individual scans. This is done by a process called registration, where the scan data's fixed relative to the dimensional survey control. However, before you can register the scan data, the initial stage of processing is to ensure that you've cleaned each individual scan, and this can be a time-consuming task, where each scan has to be removed, and any unwanted data removed, for example, removing passing cars or people. 

And once the scans are clean, you begin registering the data onto your local control network. It's achieved using one of two methods or a combination of both. The first is target registration, where physical targets are used. The design of the targets varies between laser scanner manufacturers, but they're usually spherical/hemispherical or plainer in form, with a white or grey surface to maximise laser returns. These targets are coordinated directly from your control network, or you can fix them from a previous scan. Having a minimum of three coordinated targets visible within a scan will ensure that it can be registered onto your coordinate system. 

It's also possible to register your laser scan data together without the use of any external reference targets, known as cloud to cloud. This targetless registration approach relies on suitably dense point clouds being captured with an appropriate overlap between adjacent scans to ensure sufficient matching points can be derived. Although this approach typically requires more scans to be captured, it can result in registration statistics similar to those achieved using target-based approaches. And it's also a useful approach when working in an environment where control targets are at risk of being accidentally moved or you just don't want them visible within your scan data. 

Once the registration's complete, you will have a point cloud made up of all the individual scans combined into a single model, like the example shown here. I've trimmed the front of the building off there to demonstrate the complexity that you can achieve. So you can see from the bottom image the registration process also gives us a report of a level of error within the data. In the same way as when we established the control network, this gives us the confidence that our data meets the requirements of a project and gives us the means to demonstrate and prove this accuracy to our clients. 

So after all that, you've finally got a complete registered point cloud of the subject, so now what? The point cloud in itself is generally not the desired end product. However, it does form the basis of multiple deliverables and output options. From a point cloud, it's possible to extract linework drawings or author-rectified images of sections, elevations or plans. You can produce topographical models from point clouds, all of which can be delivered as simple 2D CAD drawings, which is generally still what most end users of the data require. However, it's also possible to produce full 3D models from the data, and there are a growing number of uses from this information, from producing digital twins of structures for us at management or design work to virtual reality tours. 

And I will now pass you over to Jon, who will take you through a few case studies to demonstrate some of the laser scanning work we've undertaken. 

Jon [32:57] Thanks, Gary. I'll start us here. So I'll take you through a set of three case studies that demonstrate different aspects the ways that we might use laser scanning. The first here is Mount Grace Priory, so following on from the example that Gary just showed you. This is a property in case of the English Heritage Trust, the most complete surviving Carthusian monastery in Britain. The particular building, or area, of it that we were interested in, that English Heritage Trust were interested in, was this, which is the Priory guest house. So there's an original medieval core. It's converted into a home and extensively redeveloped in the 17th century and also undergoes an extensive remodelling in the 19th century, when it's converted into the current arts and crafts style house. 

English Heritage Trust asked us for several products. This was a pilot project to see which method, whether the static scanning or the mobile laser scanning or whatever, would provide a convenient product for them. However, for the purposes of this, I'm just going to concentrate on the static tripod-mounted laser scanning. So this is the sort of data that Gary was just talking to you about. 

So having established the control network, as Gary outlined, the scanning was undertaken at Mount Grace Priory. On the right, you see the three images, as Gary showed. We have the greyscale intensity, the intensity colour map and the RGB colour map there. So to survey this building required 159 scans. I should say that the number of scans required depends in part on the complexity of the structure and also, in no small part in a space like this, on the quantity of furniture and other stuff that's in the building when you're moving round it. Obviously, individual scans leave gaps and we have to fill those in, so the more furniture there is in there, the more difficult it can be-- or it adds scans to the process. 

So we ended up with 17 billion – more than 17 billion – points, 2.4 terabytes of data all up, so this isn't producing small data sets, and that required one person on site for two weeks. That's 100 hours of work scanning, and then the processing – this is the cleaning and registration to which Gary referred – took a further three weeks, 108 hours, before we were at the point where we have a final registered product. 

So here we see the process. So the site control is established by GNSS and then transferred to a local grid so that we can get the scanner data to work properly. Then scans were undertaken in each room plus scans to link them together. You can see the red spheres here are indicating the scan positions through the rooms, and we have link scans in doorways to make sure that scans in a corridor, for instance, can be matched with scans inside individual rooms. 

So the scanning is undertaken sequentially, so we're moving through each floor from the bottom to the top of the building. This process is made easier with newer scanners because we will often have a real-time check via a tablet or a mobile device on the progress of the scanning, and we can check our levels of completeness as we move around the structure, and we can also see if we have any immediate problems on site that we can rectify at the time. 

Once we have registered scan data set, we've got the entire building, so you can see on the top right, this is an external view. Bottom right, this is where the registered point cloud has been sliced down the middle, and this offers us an opportunity to, for example, take this point cloud into CAD for digitising long sections through the building. We can also see on the left that once we have this registered product, we can generate cross-sections through the building as well. 

So we're looking at vertical sections there. We can also use the scan data to generate plans, so once we have this combined point cloud, we can sort of slice and dice it in any way we see fit. And we can extract thin sections from that point cloud, which then allow for vector plans to be extracted, simply by tracing through the points from the point cloud in CAD. So here's an example, so I'll show you on the left. You can see a section through the point cloud. In the centre, we have some of the line work that's being digitised from that slice through the point cloud, and on the right, you can see the vector plan data as it's been extracted. 

Now, when we bring the registered point cloud into our CAD programme, typically we'll use these slices at about 5cm thick or so, and these can be moved up and down through the point cloud, so if there's particular things of interest that are not on the cutline, so a cutline through a building will typically be at about waist height so that it shows all the openings: doors, windows, etc. But if we have high-level detail, for example, a window at high level, or we're interested in a reflected ceiling plan on a particular drawing, we can nudge the slice upwards through the point cloud till it's showing the required bit of data, which we can then digitise in its correct position relative to the rest of the plan. 

The level of detail that's extracted from the point cloud to produce a drawing will depend largely on the output scale, the required output scale of the drawing – so for example, a plan that was intended for use at 1 in 100 – would have rather less detail digitised from it than one that was intended for use at 1 to 50, for example. Because of the indiscriminate nature of the data capture from a laser scanner – in other words, it'll measure anything that it can see that it's pointed at – this also allows us to extract different classes of data at different times, should the client require it and should they be shown in the data set. So for example, in this case, at a later date if they're in the point cloud, items such as electrical sockets, light fittings and such like may be added to the data set on different layers in CAD to fill out the data requirement for the client. 

Now, work is still in progress on this project. We have yet to complete the drone and terrestrial imagery, which will be used to infill the roof. You can see here, if I point in this area here, you can see that there are gaps in the roof structure here, and these directly correlate with the crenulations on the front of the building, and these go to demonstrate that laser scanning is a line-of-sight measurement method. In other words, if the scanner can't see it, then it can't be measured, so I'll be talking a little bit more about how we may integrate these data sets from other sources in our next webinar on reality capture for remote inspection. So that's a fairly complex structure, and that's the use of laser scanning to produce plans and sections and elevation drawings.  

The next structure I'm going to talk to you about is a much smaller and discreet structure. This is Garway Dovecote in Herefordshire. It's grade I listed and scheduled and constructed in the 14th century again, and it's a standalone circular structure at Dovecote in sandstone masonry. There are two entrances into it. We're looking at the south entrance here, and the reason that this was surveyed was there were concerns regarding the structural stability of the monument. There are cracks visible in the inside and also the ingress and retention of water. With that in mind, we were asked to survey it over three years in order to monitor potential structural movement. 

This photograph of the top of the structure, on the right, shows us that this concrete capping, which was installed in the 1940s – you can see the thickness of it there – is believed to be probably the root of most of these problems, and in addition to that, one of the spouts was blocked at the time this was installed. We can see part of the problem in the image on the left, which is a view directly up into the interior dome structure on the inside of the monument, and you see this dark patch, this dark circle here, is evidence of water being retained within the roof structure, so there's a lot of damp being held there, and that's one of the reasons why we surveyed it. 

If you're doing a repeat survey of any structure, as Gary's emphasised, the most important element of this is the control. So we need to have a consistent control network that isn't on the monument itself, since we suspect it's moving, so we established permanent control markers and resumption points on buildings. These included pins in the farmyard, resumption points on buildings that were not moving, as well as some monitoring points, discrete monitoring points, on the inside of the structure straddling the cracks that were visible. 

We used two methods to survey this. So the first was the 3D laser scanning, which I'll talk to you about further. This provides us with a three-dimensional record of the structure as we were able to capture it, and we also undertook a survey of these discrete monitoring points on the inside of the structure using a total station theodolite, all with reference to the same coordinate system, so all of our results can be directly compared. 

And here's a plan view of the scan setup. So each of these red dots here represents a scan position in plan view. Each of these green stars that you can see in places like this, these are all the positions of control points, some of them temporary, some of them permanent. And when we stick the registered points down together, you can see this is the network that we have. These are the matches between individual scans within that network. 

If we have another view of the survey from a sort of different perspective, in this case each of the scanned positions is denoted by one of these blue spheres, but we scanned inside the structure as well as out, and you can see that there's a rising column of scans through the centre of the structure here. So this is what those scans look like when we move inside the structure. 

So we have each of these spheres again represents one of the scanner positions, and you can see the reasons, one of the reasons, why we had to do this. Being at Dovecote, the structure is full of nest boxes for the doves. So these are all the holes here, and if we have the scanner in a single position towards the bottom of the structure, we're not going to see into those nest boxes and be able to look at the backs of those, which is an important part of looking at structural movement in this case. 

So we've scanned at different heights through the structure, and this is the method we used to do it, so we have an extending tripod, which we're raising half a metre at a time through the structure. This is matched scan to scan plus with control on the floor of the structure inside, and eventually, you can see the scanner here. We can poke it out of the top of the structure to capture the roof thanks to the open [coupler?]. 

Once we have all those scans registered together, we can use traditional methods for producing line drawings just as we did-- we are in the process of doing with Mount Grace Priory. You can see an example of one of these here, just a simple section through it showing a line drawing. These are very useful for showing, for example, roof deformation. You can see that the roof is slumping slightly, the internal structure there. We can look at deviations from the vertical in the walls, and we can break down the structure into structural elements from what we've been able to see. This is one of many sections and plans that were generated through the structure as linework drawings. 

But the other major aspect of scanning available to us is that we can perform three-dimensional comparisons of that data to see, apart from the discrete monitoring points, which we've been measuring with the total station theodolite, whether there's any other structural movement in the building that we're looking at. 

So here's an unroll, so this is-- it's, broadly speaking, a cylindrical structure. So here's an unroll of a comparison between the 2017 and the 2019 exteriors. We're missing the couple that are on top in this comparison, but you can see that we have a very high level of congruity between the two data sets. 

There are, however, some differences, notably around this doorway here, and we'll find that if we look at the scanner data images, we can see what the reason for that is. Now, as Gary said, we clean the scan data as much as we can so that we're comparing like with like, but with vegetation that's very close to the structure like that, it's almost impossible to remove all of it, so what we're seeing there actually is no difference in the structure, or no discernible difference in the structure, but we can see that there are differences where the vegetation has been. 

Here, we're looking at an unroll of the interior comparisons. This is 2017 to 2018. The results were the same for 2017 to 2019. And we can see here that again, we've got very little variation, but there are what appear to be some potentially worrying differences up in the concrete cap there, and when we look at that, the imagery from the scanner, we can see that it's changes to these ferns which are growing up there in the gap between the concrete cap roof, which is this here, this area here, and the stone roof that's been reconsolidated in this area here. 

As well as the usefulness of the imagery for pointing out the growth of the ferns, it's also extremely useful for showing demonstrably in an area that was hard to get at the amount of water retention on one side of the structure there, so extremely useful. And so here we are looking at a three-dimensional view of the exterior, so when we'd finished the comparisons of the epochs, you'll be pleased to know that there weren't any discernible structural movements of any worrying significance. The difference is we have our vegetation, which is evidenced down here in this exterior view. 

The only thing that really did move, the part of the structure that really did move was the waterspout, which you can see here. I don't know if any of you remember, but in the UK here, the summer of 2018 was extremely hot, and this is a largely unsupported lead spout. There's a small piece of stone there holding the base of it in place, and you can see from our data that that is showing up as a bright red spot. This moved by 7 or 8cm over that summer, so it was both a vindication of the fact that the survey control was working correctly and also that we were spotting differences where they occurred. 

The other major difference that we had on the insides was swirls of material that were moving and weren't relative to the cracks and these turned out to be, and I don't know if you can see this here, but these are cobwebs which are very thick and laden with dust, and any movement of air through the structure was causing these to move, and they were being picked up by the scanner as differences. So we can, again, check the imagery and make sure that what we were looking correlated with what we were expecting to find there. 

I'll move on from Garway to a survey at a different scale. We're looking at very small stuff now, so we're using a close-range scanner. In this case, this works-- a structured light scanner works by projecting a light pattern onto the surface of the subject, and we used-- so that comes through a projector, which is in the centre of it here, and then we used two cameras, stereo cameras, to pick up three-dimensional variations on the surface. They're very short range, typically from 10cm up to about 3 metres. Very high accuracy. We're in the sort of low microns area there. In industrial metrology, they're very popular for scanning, for example, car parts or areas of metal, where they use blue light because they're not interested in colour. In heritage, as Paul mentioned, we use a white light projection because that gives us colour information as well. 

There are some major disadvantages. Main one is they require mains power and very stable placement. Because this is measuring so accurately and precisely, any movement will result in messed up scan data, so we can't use them, for example, on a suspended floor because if somebody walks across the floorboards, that will be enough to misalign the scanner. They're also expensive – £20,000+ for a scanner. 

So the example I'm going to talk to you about is Richmond Cell Block at Richmond Castle in North Yorkshire, and the cell blocks here were used for housing after the First World War, during and after, conscientious objectors, but also a wide variety of other people were imprisoned there. The joy of this place is that the limewashed cell walls are absolutely covered in graffiti, many layers of them, and we were there courtesy of English Heritage Trust to do some close-range scanning for the location of some paint sampling points. You've got some little arrows on the upper drawing, showing where they were, and also for monitoring of change. We scanned this two or three years previously with a different close-range scanner, so we were looking to compare the results of these different epochs of scanning. 

This is one of the sample areas, and you can see the variety in-- at least part of the variety of richness and fun of the graffiti in there. And I shall zoom in a little bit closer. So one of our paint sample locations was actually just off this scan down here, but we're looking at not a picture, not a photograph, of the surface of that wall, but a three-dimensional model. There are some occasions where, for example, the surface characteristics of something may be more important or interesting to you than the colour information here. And in order to look at this, we can strip off the colour information, and I'm hoping that you can see there that we're seeing different layers of the plaster and areas of cracking without the distraction of the colour. 

If I switch back to the earlier slide, you can also see how some of the graffiti has been incised or drawn with a much heavier hand than others. So this cursive script in this area here doesn't show up at all in the surface model, whereas HWAR, which you'll be able to see, clearly does. So we can see the scratching of HWAR here, whereas the cursive script here has left no discernible surface trace at all. 

One of the other advantages of a three-dimensional model is that we can take measurements of it, so in this area, where there's an accretion of plaster here, we've been able to measure from this. So you can see on the right here, these three blobs here correspond to these blobs visible on the surface in the image here, and we're able to take measurements there. I don't know if you can see that on this, but it's 0.423mm. 

As well as looking at the surface as a point in time, we can also, as I've said before, make comparisons with earlier epochs of scanning to show areas of surface change, so that's areas of both accretion, where, for example, the surface may be bubbling up due to water and salts underneath it and also areas of loss, so we can see where stuff's just fallen away. 

That's the end of the case studies, so I'll just now point you at the poll question three. So would any of these outputs and analyses be useful to you in your work? Well, that's good to see. I imagined 'occasionally' would be the most popular there. OK. I'll hand you back to Paul now. 

Paul [57:36] Thank you very much, Jon, and thank you very much, Gary, before that. We're not one minute before 2 o'clock, so we are going to go a little bit over. It should only be about five minutes. But I just want to sort of summarise this. 

If you think back to the image I showed of the Cyrex and the RIEGL scanners back in the late 1990s, admittedly, that's over 20/25 years ago. They have developed an awful lot since they first arrived on the scene for heritage, and if you think about the sort of heritage applications, typically they can be used on, as Jon has suitably described and Gary before him, on 2 and 3D surfaces. The technology is getting faster all the time. That means that the time to undertake the surveys is coming down, and I know there's been a question that's been raised that's generated a lot of chat regarding the cost of the survey. So as scan times do come down, then the time needed to be on site comes down as well. 

However, the scanners, they are very fast these days, and as I say, the latest scanner that we've started using can do a full 3D scan to the 3–4mm accuracy that Gary highlighted within about two minutes, so it means that the efficiency of the scan capture on site is increasing. But that is also having a knock-on effect for the data sizes, so it does mean a lot more data to process, to pass on to the client, and then there's the problem at the end of how do you archive all of that? Maybe that's another webinar entirely. 

Later scanning has the benefit of being able to generate 3D-pointed data in the field, so effectively, you can leave site in the knowledge that you've captured all your data, it's all registered together and you can then post-process it to generate your outputs. The mobile solutions, they are allowing data capture on the move, but I am going to be a little bit controversial here, and I've been a little bit disappointed in the developments from the mobile scanner manufacturers. I was expecting higher-resolution sort of scanning solutions to be available at the moment, but I think the terrestrial scanners, they're actually catching up in terms of being able to capture entire sites in a short time. 

Another interesting fact for me is that as well as 3D point data and RGB image data being captured, we've now got thermal, and I know that there are universities out there. I think we've got one representative on the webinar today from Nottingham Trent University, who are doing extraordinary research on other multi and hyperspectral sensors that could potentially be added to a terrestrial scanner, thereby increasing the amount of data that could be captured and to enable classification of the surface that is recorded. 

As I say, one of the downsides to laser scanning is that they are generating very, very large data files, and they are still difficult for some people to use without high-end computers and specialist viewing softwares. Manufacturers like Leica, they have JetStream technology, which means that you can view entire data sets for a whole structure in real time, but you still need to access the data to be able to sort of process it and deliver that sort of end result. 

The scanners are expensive. I don't see them coming down in price any time soon. You still need sophisticated software what I call a useable output. I started a question as regards the automated drawing extraction tools that are out there. They've been out there for quite a few years now, mainly in the process plant, where they're used pretty effectively. 

For heritage, I've been slightly underwhelmed by their ability to generate a useable automated drawing without needing a lot of post-processing, editing afterwards. And then, sort of highlighted there, line drawings still require manual digitisation. That's exactly what I've just been talking about, although, as has been shown in a few of Gary's slides, the orthorectified outputs, they are an alternative to a drawing that's a lot quicker to generate, and they have the benefit of showing you the full service rather than an interpretation of that surface as a drawing would generate. 

So final few slides. We're talking about remote inspection and the ability to actually capture data that enables you to then go away and then analyse that off site. Laser scanning provides a very good solution for that, and this is a project we've been working on recently. We mentioned it, I think, in the drone webinar. Largest timber structure in Europe, a lot of faces that are hard to actually reach. Laser scanning can do a very good job there in picking up detail to aid conservation repair and maintenance, but there are still gaps in that that other technologies will probably need it to fill them in. 

Generating surveys is one thing, but it needs to be for a purpose, and typically that purpose is to provide an output for, for instance, conservation repair maintenance works to be either planned, specified or even undertaken. Laser scanning is a technology that is very, very good for generating the base data from which such outputs can be created, and there we've got an example from a small project we did at Woodhorn Mining Museum on a structure that's on the heritage at risk register, and the outputs were purely based on the intensity output from the scan data. Very useful data in its own right. Didn't need any digitisation, and it's sort of a straight extraction from the scan data. 

Both Jon and Gary have the title 'analyst' in their job title, and that indicates that geospatial data has a use for analytical purposes. So once captured, yes, you can generate geometric shapes to help record the structure, but there's also a lot of information there, particularly then the surface data that's collected that can then be analysed by other experts, for instance, archaeological experts as shown in these two example slides. One of them is about a small research project that our previous placement student [Leigh Sue?] is now doing extraordinary work on behalf of Historic Environment Scotland. Did for us over at Carlisle Castle, where she used laser scan data on a historic doorway to extract out the carvings that were known about but hadn't been fully documented before, and it's quite an extraordinary piece of work that used, effectively, one or two scans from a Leica P40 laser scanner. 

The other one on the right is the Stonehenge survey I mentioned earlier, and in terms of heritage projects, that's a very significant piece of work because it highlighted all the archaeological information that a laser scan survey can enable extraction. But also, the end uses of that data and that survey that was captured in 2011/12 is still being used. The data sets are still being used, reanalysed, repurposed, generate new Visitor Centre presentations in the actual site. So it might cost a lot to actually generate the base survey in some people's eyes, but if you think long term, that sort of benefit, that cost, really, sort of has beneficial applications further down the line. 

There I'm talking about presentation, so there we've got the Stonehenge Visitor Centre itself. If anyone's been there, in between lockdowns, you'll have seen this, and this is just one 360 image that's generated from that 3D laser scan survey that was done about ten years ago. The image on the top is for another small heritage at risk project that we undertook for our colleagues in the Yorkshire region of Historic England, and there just simple slices through the scanned data highlights the structural issues that this building was suffering from. And it helped the owners to actually come round to the idea of refurbishing it and effectively taking it off the heritage at risk register. So that's simply a slice through the data, but it's very powerful in what it can actually provide. 

Then the final bullet point, final piece in the webinar, is building information modelling, or BIM. Within the geospatial survey world, we often hear about scan-to-BIM modelling, so in other words, it's capturing laser scan data for buildings and structures and then going through a process to generate a BIM model from that, or it could sometimes be called a BIM-ready model because it doesn't actually have the full process applied to it. 

So here is where I'm going to stop the webinar and actually sort of bring in the final poll because I want to just gauge people's reactions to BIM. Some of you may never have heard of BIM – 'Don't use BIM', that's gone straight up there – but I wanted to sort of know if anyone's already using scan to BIM and whether there is interest in a further webinar on BIM for heritage within the Technical Tuesday series. I'll let that run a little bit longer. We're getting 50% of people don't use BIM. Interesting. I'm pleased to see there's 35% of you out there that would be keen on a BIM for heritage webinar. 

So if we close the poll there, Alice, I'll take that away and speak with the people that are running the webinar series and see if we can put something on to do with BIM for heritage that will maybe address some of the issues that the people who don't use BIM have on its use. 

So apologies for going ten minutes over. Thank you very much for staying with us. I've seen we've got 144 people still here, so well done to you all for staying with us. My contact details are there. We had a few questions asked. I don't know if Alice has managed to bring them across. She was going to put them in the next section. 

Alice [01:09:57] Hi. Hi, Paul. Actually, we had David and Gary answering a lot of the questions, and Kerry, so we had most of the questions answered. We had one or two that weren't, two very quick ones that I could ask you that weren't actually answered... 

Paul [01:10:12] Please, yeah. 

Alice [01:10:14] ...or may have been, but I may have missed them. So the first one that I could find that wasn't answered was from Chris Brook. 'Are the areas specified RMS or absolute?' I don't know if there was more context to that question, but the chat has been going bananas, which is excellent. If we can't answer that, I can go into the next one. 

Paul [01:10:38] Yeah. I think I'd probably pass that one over to Gary because he was the only one of us, I think, who was talking about errors. Are you on the call still, Gary? 

Gary [01:10:47] Yeah, I just answered that one. The GNSS that I quoted was RMS, and the laser scanner was the absolute 3D positional accuracy. If you're really interested, you can google product specifications and get all the numbers. It's all very exciting. [crosstalk] 

Paul [01:11:08] Thanks for the question, Chris. 

Alice [01:11:11] And then the last one that we haven't answer because, as I said, they were very hot on answering the questions, which was really great because obviously we've gone a little bit over time was from Ed Bristow, was-- I know you meant-- His question was 'What sort of training is available to get into this field of LiDAR scanning, etc.?' And I know you briefly mentioned that there's somebody from the university, and they're starting to run more courses, so if you could elaborate maybe. 

Paul [01:11:38] Right, well, if we'd have been doing these webinars about five years ago, we could have offered them our survey summer school that we used to undertake on an annual basis. That was back in the day where we had face-to-face training. That just seems so long ago. I'm not sure whether we'll be continuing that, given we're not sort of heading down a digital road. We ourselves, we're doing the webinars, so I think we'll be increasingly doing this sort of presentation in the future. 

There are the academic institutes like Nottingham University, Nottingham Trent, UCL down in London, Newcastle University, which was where I went. They do courses, but they're aimed at sort of the higher end of the discipline, really. If anyone's interested, I'd approach the TSA, The Survey Association, that I mentioned right at the very beginning. I know that we have Rory Stanbridge on the line. He's probably not able to answer to this, but the TSA has the Survey School, which is based down in Worcester, and it's able to offer training within all of the geospatial disciplines, so I'd direct people to the TSA, to people like Rory and Rachel, who is secretary general at the TSA, and I'm sure they'd be delighted to advise you on that.  

We've also got new research opportunities with collaborative doctoral training partnerships. One in particular is between Nottingham and Newcastle universities that was recently launched, and it's looking at fairly advanced geospatial research, not just in heritage buildings but across the board into climate change, monitoring of structures, spatial analysis of how structures perform, all based around geospatial data that is becoming more accessible through the work of the Geospatial Commission. 

So a bit of a politician's answer there, rambling on maybe not answering the question directly, but there's a whole host of opportunities out there. Rory's just said, 'Was just going to suggest the survey school'. Thank you very much, Rory. Let's say that's the first port of call for that sort of question. 

Any more, Alice? 

Alice [01:14:29] OK. I've got two left. I know that [Suki Ri?] had some questions answered, but the last one is 'Has anyone used LiDAR sensor on the iPhone Pro 12, and any thoughts on it?' 

Paul [01:14:45] I'd be very interested to hear, myself. I do like a bit of technology. I don't use Apple products myself, so I haven't any experience. I would say that we have seen scanning sensors on phones before. It's not new. They've been around for several years, and if you look at our 3D laser scanning for heritage guidance, you'll see a reference to some of them in that. They're not designed for the sort of scanning that Jon and Gary have been talking about. 

They're more aimed at visualisation. For instance, if you wanted to check if a new sofa fitted in your room, you could do a 3D sort of scan of your room on your iPhone, and then you'll see augmented reality sofas and chairs there. They're also good for, like, gaming applications. I don't think they're going to be useful for sort of high-end survey purposes, but I'll put my hand up; I haven't used one, so if anyone can enlighten us on that. Maybe that's a piece of research for the likes of Neil at the Geospatial Systems CDT at Nottingham and Newcastle. 

Alice [01:16:14] And then our last one – and thank you, everybody who's still with us. We still have the majority of the people – is from Charles Richards, and I think this is the last one we have time for. 'Can 3D laser scanning be used to accurately date buildings?' I know there was a question from Deidre about can the laser tell the difference between different eras, so different timelines in a building, and I think David Andrews answered that quite well, so I think it's kind of an extension of that question. 

Paul [01:16:42] All I'd say on that, Alice, is a 3D laser scanner itself won't be able to answer that question. I always say that 3D laser scanners, they are quite dumb instruments because they have the idea of the generating 3D data. It's how that data is analysed that is the key to maybe answering that question, and if you're able to capture high-quality, real-world colour alongside potential other spectral data, which is why I get so excited when I hear about what's being done at Nottingham Trent University. That could then potentially lead to semi-automated analysis, potential dating as well, but there's always something else to do with a laser scanner. Us geospatial surveyors, we're never happy with what's being provided. 

Alice [01:17:36] There's always improvements to be made. 

Paul [01:17:39] There's always improvements. So maybe if we finish the webinar on that point. 

Alice [01:17:43] Yes, there we go. Thank you so much, Paul, Jon and Gary. Thank you so much for a real enlightening webinar today, and thank you to all our guests. The chat was, as I said, bananas. There were lots of good questions, and thank you to David Andrews, who stepped in and was answering questions. I know Gary, you also answered a few questions, but it was faster than the eye could see, the majority of the time, so if I haven’t thanked anyone, please know that I am--