Home Articles Algorithms for near real time image analysis are today’s need

Algorithms for near real time image analysis are today’s need

Robert Moses

Robert Moses
CEO & President
PCI Geomatics, Canada

Around 1985-86 we entered the image processing and remote sensing market space as a result of Canadian government’s project related to the working of satellite remote sensing applications on a parallel processor. In those days we were in the middle of the cold war and the government was trying to access crop yields of Russia, China and other Asian countries, to try and predict whether there was food shortage, etc. Today we use the same technology for the commodities markets.

This eventually resulted in development of the PCI Geomatica software suite, which we were selling with package deals at USD 50,000. Later, with rapidly developing IT industry and lowering software prices, the similar packages were sold at USD 5000, which in turn led to strategic decision making with a business point of view.

We realised we had to expand but could not compete ESRI and other well established vector based GIS companies. At the time, satellites sector was growing at a very fast rate, both in commercial and secret government national assets all over the world. It then occurred to us, about 7-8 years ago, that there will be a bottleneck in the industry both in terms of processing of the massive collection of data and its analysis for user specific applications. In order to succeed in our endeavour we had to convince the government departments about our presumptions and the need for image centric software and solutions. We successfully convinced three departments – Defence, Homeland and Security and Natural Resources. They then sponsored us for a large project within Canada and a technology partnership fund was allocated to create the building blocks for an image centric scriptable type of systems.

Once we realised that the image centric market was an open space and if we could possess it correctly and create a system around it much like the systems around the vector methodologies, we could help battle the bottlenecks like processing and analysis of the image data.

In the past, basically we used vector methodologies with an image backdrop, but I recommended the use of image technology with a vector backdrop; to use vector data and attributes that we gather to help process the image, not vice versa.

We developed certain applications that were much easier to do because of raster nature of the data and also easier to automate the process in certain respects. We could process all that data which was natural in the raster image world and very difficult in some cases in the vector domain.

Nowadays, not only the satellites but UAV’s (Unmanned Aerial Vehicles), LEO (Low Earth Orbiting) constellations, HALE (High Altitude Long Endurance) vehicles and even digital cameras are all creating terabytes and petabytes even zetabyes of image information. And very soon we will not be able to keep up with the high quality and quantity of image data in flow. Our business case is thus cheaper and faster because it is much scalable and better because I believe that this technology from satellite is much richer than vector maps. There are many modelled and layered information available but the satellite image has it all in one place – spatial features like road and buildings, water bodies and forests, etc., even spectral data from multi/hyper-spectral satellites which are fairly temporal in nature as well.

We were one of the six founding member of OGC. We could have never achieved this limited success unless we pushed for OGC national standards. It is difficult for small companies to have and maintain different formats. Small companies would never be able to sell in Canada or US or India or Europe, unless they stick to OGC standards in order to make big international successes. The company decreases the cost of creating the software and decreases the barriers to penetration internationally. Now from government’s perspective, if you have only 2-3 monolithic companies with closed proprietor’s standards, the cost of the systems goes way up. And that is why US government has mandated on all procurements to be OGC compliant open standards. The idea basically is to have a flexible, scriptable/customisable and interoperable platform based on a centralised server and a data management system with geospatial capabilities.

We as an international organisation need to work together to solve issues of global warming, climate change, terrorism and wars that we are causing to our own planet. All these issues are in a way geospatial and to deal with it efficiently we need a system that is interoperable. I believe in coming 3-5 years we would reach, if not ideal, at least a near ideal situation in interoperability, even in India.

No one country has enough skilled people to keep pace with the processing of image information. By eliminating the requirement for pre-processing of the image information using PCI software, we believe that we are creating opportunities for people to better allocate time and resources to research and fill vacancies associated with analysis of the giant amount of image information that will be available to us over the next five years. We need to have researchers who will help develop algorithms for improving the automation process and making user-defined data available in near real- time basis. The researchers should focus their learning towards spatial analysis techniques.

We have to shift our focus to much higher quality job which is closer to decision making process because in the end this information will be made available to the decision making agencies and individuals for emergency response or anything of that nature.

We are going to start operations in India that will provide PCI software packages along with solutions for Indian private and government sectors. We are working towards tieups with top 5 to 10 Indian companies that truly own the market to provide them with underlying image-centric technology and allow them to script it according to their requirements. We are also planning to work with some governments like Punjab to help them develop agricultural information systems. It will be hard to sell them the systems, hence we are also thinking of providing them with the information and solutions.

We are witnessing image data explosion and extracting information would be lesser of activity and people would be working directly on these imageries as we do with the vector-centric Geomatics solutions.

There is too much image information available and it is going to be wasted unless we switch our paradigm to an image-centric paradigm. It is our vision that PCI becomes a ‘platform image-centric technology provider’ to all the corporations, governments and individuals and help them process their image data to extract images for further analysis that will support the end use.