G-Video: The new way to capture and manage High resolution spatial dataGonzalo Garcia OlagortaCEO of Geovirtual S.L. Al Amjad L.L.C, UAE Email: info@alamjadco.com Sometimes, when we join different technologies in a smart way, we get something out of them with features that remind us of the parent technologies but with its own entity. We can say then that something new has been born. This is the case of G-Video that stands for geo-referenced video, the result of joining GPS data to digital video. I will explain the basics of the technology to encode the coordinates of the position of a video camera along a path, how to export them as a vector and incorporate that vector into a 3D scenario. We will also describe all possible uses of this technology, its benefits and its applications. Our conclusion will be that G-Video is probably the fastest and cheapest way to capture and manage high resolution spatial data concerning linear assets, such as pipelines, roads, electricity networks, railways and others. I have dedicated a good part of my life to developing innovative graphical languages for the description of the territory for the public at large. I started designing books and maps for tourist and coastal navigation where aerial photography played an important role. Later I incorporated in the projects the techniques of 3D cartography and during the last 9 years I am fully committed to software systems for what we call: dynamic 3D cartography. In short, systems similar to Google Earth but for professional use. Looking backwards in my curriculum, I can see very clearly that I have always felt very much attached to analogue representations of reality, mainly by means of aerial photography, both oblique and vertical. And I have been faithful to this resource because I have always found it convenient for being easy to understand, convenient for being natural. At the very beginning we started our projects flying with helicopters and small planes, using simple cameras to capture oblique snapshots. On top of them we redraw cartographic masks with vectors and symbols. Soon this procedure to represent the territory became insufficient to cover bigger areas, and that took us to using orthophotomaps, both from aerial and satellite images, on top of digital elevation models. It was a matter of time going from 3D still images to attempting to fly over 3D scenarios. In 1997 we wrote the first software application that let us do smooth flights with a reasonable performance in frames per second terms. Since then, it has been a long way in tuning-up the software by introducing new features and characteristics, with the target in mind of making consulting geo-spatial data, a pleasant and rewarding experience. Getting data has always been the constant that has accompanied us along this technological evolution. Now that our work requires managing immense datasets in order to recreate realistic 3D environments, we miss the good old times when data was easier to collect and closer to our reach. Projects now require the use of very high resolution ortho-images; otherwise clients and final users are disappointed with the result of our work when they fly close to the ground, at a distance where the brain expects to find volumes of elements which geometry has not been included on the digital elevation model. This happens with buildings, trees and human assets. Actually they are right, taking decisions require the best possible information available. The problem is that that data is not always there. It was only a couple of months ago that I was introducing a project that we had developed for the cartography department of a national gas company. We had built a 3D scenario of the whole country and linked their Smallworld GIS to our software that allows smooth flights over the territory. As it was a very large country, we had used medium resolution satellite images. The result was quite good, as pipelines could be clearly seen spread all over the country. When looking at the big pipes from high over the ground, the view was good enough to identify the target, but when we were going closer to the pipelines, the view was, evidently, strongly pixeled. Client suggested then to acquire very high resolution satellite images of a corridor of 1 km buffer along the 5.000 km pipeline. A very expensive and hardly effective solution. There is however a much better way to solve that problem. A way by which we can get extremely high resolution spatial data fast and cheap. We call it G-Video. Using videos taken from ground or aerial vehicles has been since long recognised a practical mean to analyse assets wide spread over the territory. The unsolved problem has always been how to incorporate all that data onto a GIS in an effective way that takes profit of the cartographic discipline of the system. I am going to explain now the approach of my company to this subject, the benefits and applications of the system and how we foresee the evolution of the technology. What is G-Video? G-Video is a file format that contains encoded the following type of data:
Capturing G-Video There are many possible combinations of hardware to capture G-Video. The one I propose here adapts very well to the requirements of most of the projects that we usually carry on for professional markets. The following diagram shows the hardware architecture: 
Viewing and using G-Video Once the data is processed, it becomes a file that can be directly imported in our 3D GIS system. Inside the 3D scenario it is shown as a vector that represents de track along which the camera has moved when capturing the video. When the user clicks on any point of the vector, a video player appears starting the video at the photogram that corresponds with the position of the camera in that place. Simultaneously with the playing of the video, an arrow on the vector moves along the path pointing at the direction that the camera is heading. The user has the possibility of accelerate, play backwards or stop the video. All data referring to the position of the camera at every moment is screened on the 3D viewer. The user interface also allows the possibility of reproducing the video in “Slave camera” mode. That means that the virtual camera of the 3D viewer moves synchronised with the position of the video camera shown on the G-Video vector. So the user can follow the track that the camera recorded from any position, and even adopt the point of view of the people who recorded the video. It has to be clearly understood that even though the G-Video faithfully captures the spatial data corresponding to a linear asset, its vector representation on the 3D scenario does not necessarily correspond to it. Actually, it almost never does. This is because there is always a distance between the camera and the object that the camera is pointing at. So we will see on the 3D scenario two lines. One is the raster drawing of the linear asset shown on the orthophoto (a pipeline, a road, a river) that is wrapped on top of the DTM. The other is the vector line that corresponds to the G-Video. Sometimes, these two lines may overlap apparently; this is when the video camera is on board a ground vehicle and pointing in the same direction of the path. In this case, the photogram captured and attached to every point of the vector will correspond with a point that will be represented some distance ahead. This is obvious. It happens when, for instance, we are recording a road with a video camera pointing ahead. The sign posts that are shown in a particular photogram are actually some meters ahead than the point of the vector where the arrow is located. When the camera is onboard an aerial vehicle, the two lines are clearly distinguished, as the G-Video vector is situated above the ground, at some distance of the linear asset that is recording. For instance, this is the case when capturing a pipeline from a helicopter with a camera that is pointing downwards and sideway of the direction of the path. In this case, the point of the pipeline recorded in a particular photogram, a valve for instance, will be found at the intersection of the arrow of the G-Video vector with the pipeline shown on the orthophoto. Evolution of the technology We foresee that in the next future we will be able to get richer visualisations with new and more sophisticated features of our 3D software system. The data of the position, movements and field angle of the camera that we record when capturing G-Video, will be used to transform the simple arrow that now runs along the vector, in a 3D kind of light beam that will intersect with the 3D scenario, leaving a sort of trail on the territory. Being this trail a vector itself, it could be consulted by clicking directly on it. The nice thing of this feature is that the user will be able to click directly on the point of the linear asset that he wants to recall and see in the video. Another way of getting the same result is by attaching a laser range finder to the video camera. The vector that is captured by the laser corresponds with the object shown at the centre of the photogram. We can say then that the laser beam leaves a trail on the ground, creating thus a second vector, again, representing the territory shown in the centre of the photogram. There is still an even more complex possibility. We could also reproject the video over the ground and replace on that trail the orthophoto by the video itself. This will give a very realistic view of the 3D scenario, but only when the virtual camera is looking at it from the position and in the direction that the real camera was pointing. Applications and benefits of G-Video As you can imagine by the description that I have made of how to use the technology, G-Video is an extremely valuable resource to capture geo-spatial data in many fields and for many different applications. We are here in Dubai, in the centre of a region where there are hundreds of thousands of linear assets, such as gas and oil pipelines that carry a very great part of all the energy that the world is using. This network system is only recognised by most of the GIS of the companies and public institutions that manage them, as simple vectors, lines spread all over the territory. This is in fact the symbolic representation that stands for real pipes, valves, connexions…It is a convenient system for many purposes but we have to recognise that, being symbolic, it is a very simplistic representation of what is really going on out there. The two big drawbacks of symbolic interfaces are:
Besides, it is also a very fast and inexpensive way of getting geo-spatial data. Processing time is a matter of hours from capturing to viewing, being even possible to implement an on-the-fly system that shows, in a fleet control room, video streamed data from ground or aerial vehicles. With these characteristics in mind, I would recommend G-Video systems for companies or public agencies that operate manage or maintain assets or territories that can be referred to spatial corridors, such as Oil, gas and water pipelines Electrical networks Roads Railways Rivers Coastline Streets And similar The tasks that are better fulfilled with G-video are: Monitoring over time Presentations or briefings Public awareness Training Identification Inspection Disaster Assessment Site surveying Construction phase reporting Environmental compliance Asset inventory Maintenance planning Damage assessment Land survey Fire line mapping Forest fire reporting Erosion control Land use planning Environmental impact assessment Road construction And others It is not my intention to tell you in which way you can benefit from using this technology, as you will probably know by now, how and where you will use it or recommend it. I have just tried to raise your awareness of the basic traits of this useful technology. I have also to mention, that even though all my comments have referred to the using of G-Video technology in a 3D environment, being 3D the technology that my company produces, all the applications and most of the benefits are also met by 2D GIS. The strong point of 3D software in this case and as we use it in our projects, is that it is also a natural approach to geo-data through an analogue interface, which makes both technologies absolutely synergic. The user flies freely over a photographic 3D scenario and clicks on a vector that runs along a linear asset to see a high quality video of the place, the experience of being there. |