In 1973 I encountered Terrestrial or Close Range Photogrammetry when I was privileged to field test the Wild P32 camera that had just been introduced into the UK.  The first project this was used for was to map the façade of a bridge in Lancashire for a road-widening scheme.  Terrestrial Photogrammetry lent itself as the solution to this task because the landowner on the north side of the river exercised his riparian rights and forbade the crossing of the river centre-line onto his property.  However the next task that the Wild P32 was used for was of much greater interest.

A baronial mansion, Annesley Hall north of Nottingham, with a cellar dating back to Norman times, was to be mapped in its entirety.  The Wild P32 was to be used to map all the external facades of the main building and the stables by Terrestrial Photogrammetry.  The fact that the Wild P32 was mounted on the top of a Wild T2 theodolite meant that the position and orientation of each and every photograph could be recorded and that the camera could also be accurately positioned at specific locations for taking the photographs.  I carefully marked out the camera locations and took considerable effort to ensure that the camera was at the correct 3D location for each photograph as is evident from the photograph.  I also placed Photo Control Points on the surfaces being mapped because it was not standard practice to know the location and orientation of the camera at the time of photography.  These were positioned in the ideal locations for the photography because I knew exactly the area covered by each photograph.  In the event, the Photogrammetric Operators ignored the information about the location and orientation of the photographs and used the Control Points as in conventional aerial photography turned on its side.  However, the information relating to the camera location and orientation meant that the facades could have been mapped from the photographs by Photographic Intersection without using a Photogrammetric solution.  With today’s technology, computers and digital images, this could be a more expedient and accurate method, but these things were still in the future at that time.

The concept of Photographic Intersection predates the invention of photography itself.  Surveyors produced drawings in the field by projecting an image onto a surface using a lens, then tracing in that image.  Maps were then produced from these images in much the same way as we use photographs for Terrestrial Photogrammetry today.  This process was given a considerable boost by the invention of the P30 Photo Theodolite by Heinrich Wild of Heerbrugg in around 1922, half a century before the Wild P32 camera arrived in the UK.  The theodolite portion of the Wild P30 incorporated the elements of the legendary Wild T2, also invented by Heinrich Wild in the early 1920s, which was a major innovation in the world of surveying instruments.

The Wild P30 Photo Theodolite  -  photograph reproduced with the kind permission of the Virtual Museum of  'UNSW Surveying Instrument Collection'.

The Wild T2 1" theodolite invented by Heinrich Wild in the early 1920s.

I first came up with the idea of using a camera fitted with a fish-eye lens.  By placing the camera on its back at a known location, the resulting image would have the points of interest radial from that point.  Unfortunately, the nature of a fish-eye lens meant that the part of the image that was of interest was in a narrow band around the circular edge of the image, and were therefore difficult to distinguish.  However, the principle was there and this lead to my invention of the conical mirror or conical prism.   This provided a distorted image, but one in which all the points of detail were radial from the centre of the image.  By taking two or three such images in a room at known co-ordinates these images could be used to plot the room by Photographic Intersection.  Using a photographic solution meant that all the detail on the room was capture at one time so that if additional detail needed to be plotted, this could be done without having to revisit the site.  I was granted a Patent for this invention in 1975, but being of very limited funds I let this lapse when the laws were changed in the early 1980s and the annual fees increased dramatically.  Within a year or two of letting the Patent lapse I noticed that Kodak Carousel Projectors were using conical prisms so that the infra red signal to change the slides could be detected from any part of the room.  Conical mirrors are used to today in laser levelling devices and perhaps most galling of all is that Trimble (Geodimeter) uses a conical mirror for their 360° active detector for their “robotic” Total Stations.  The digital computer was still in the future and would have been a great asset in developing this technique, but actually presents more suitable solutions.

A method of Photographic Intersection did solve a real problem at the beginning of the 1980s.  Due to a tunnel collapse in Glasgow with fatalities, it became desirable to know if there were any distortions in Railway Tunnels.  Stereo Photogrammetry was considered a solution but it was soon discovered that there was a huge difference between measuring images that the principal rays of the cameras were perpendicular to, as in aerial photogrammetry, and measuring images where the subject was parallel to the principal rays, such as in a Railway Tunnel.  This was not helped by the fact that the tunnel walls were also black from years of soot and grime.  Ian Waite had used a flash gun sandwiched between two sheets of plywood to illuminate sections to determine overburden when tunnelling for the Dinorwick Pumped Storage Scheme in Wales.  It was considered that if such a “light line” could be introduced into the stereo images then the photogrammetric solution would be viable.  Two Companies addressed this problem, Photarc Surveys in Harrogate, Yorkshire and BKS Surveys in Coleraine, Northern Ireland.  The BKS team consisted of David Stevens and me. 

Theodolite Intersection still provides the most accurate method of measuring objects remotely.  We can expect sub-millimetre accuracy over ranges in excess of 100 metres when using top-of-the-range Instruments such as a Leica TPS1201.  The technique also provides a feel for the quality of the observations.  However, the data capture process in the field is time consuming as each point must be observed from a minimum of two, and preferably three, Instrument Stations.  Mistakes can easily be made in identifying the same point of interest from the different Stations.  The information you take away from the field is only the discrete measured points.  Additional data will require another site visit, but the site could have changed, or may even have ceased to exist.

Reflectorless Measurement also requires less time in the field than Theodolite Intersection as each point needs only to be measured from a single Station, but is still considerably more time consuming than Photogrammetry.  Accuracy is better than with Photogrammetry but not as good as with Theodolite Intersection and is more dependent on the accuracy of the Reflectorless EDM than angular accuracy of the Instrument.  The accuracy of the Reflectorless EDM is affected by the nature of the surface being measured to, and there is no information as to the quality of the individual measurements.  Once again, the information you take away from the field is only the discrete measured points.  The range is restricted by the Reflectorless EDM.

Scanning with a system such as Leica Geosystems HDS creates a large “point cloud” in a very short space of time, so the field time for data acquisition is very short when compared with Theodolite Intersection and Reflectorless Measurement.  The density of the ”point cloud” is determined at the time of measurement so missed discrete points or areas would require revisiting the site, although the “real time” display of the data should enable the user to ensure that everything of interest is covered at the time of survey.  The equipment is however still relatively large, heavy and expensive when compared with the other techniques, but is the future as far as collecting 3D data to the precision required by Surveyors and Engineers.

Terrestrial Photogrammetry requires much less field time than Theodolite Intersection, but accuracy suffers as the camera locations are reconstructed from Photo Control Points in the images, and these points have to be measured using another technique, such as Theodolite Intersection or Reflectorless Measurement.  A considerable benefit is that a huge amount of information is captured in the images and this can be extracted at any time in the future.  If the subject only appears in a single photograph it can often still be measured using the geometry of the structure and the single image.

Photographic Intersection involves little time in the field to capture a large amount of data that can then be extracted as and when required.  Accuracy will not be quite as high as for Theodolite Intersection, but will probably be better than with Reflectorless Measurement, depending on the pixel resolution.  Accuracy will certainly be better than with Stereo Photogrammetry and probably better than with a Scanning system.  This could be the most cost-effective solution for a huge number of subjects.  I am promoting Photographic Intersection as a low cost solution for capturing 3D data to a high order of accuracy because the equipment required is easily obtained at little cost and the mathematics involved is straight forward trigonometry.  However, it is my opinion that the best method for serious 3D data capture is HDS (High Definition Surveying).