Bob's Adventures in Wireless and Video Headline Animator

Sunday, January 23, 2011

I thought I saw the coolest VMS in the World

I hold in my laptop a demonstration of the most exciting Video Management innovation I have seen. Or at least that I had seen up until three days ago.

I love Google Earth. It is fun, intuitive, fast, and provides access to a mind boggling amount of knowledge in an easy to understand user interface. If only there were a way to link in real time video streaming from cameras and overlay them on the Google Earth map, that would be the coolest VMS interface on the planet, or Earth.

That is exactly what some really smart people from academia have created: a technology which is nothing short of awe inspiring. Using distributed computing, lots of video algorithms and some leading edge modeling tools they are able to dynamically overlay real time video onto two and three dimensional models.

The result is...well take a look here

The system scales to thousands of cameras and can contain meta data which can create alarms and alerts based on camera analytics that appear on the map. It allows you to not only view all of your cameras in a context that is immediately obvious, but you can map data into the map as well. So instead of seeing all of the gas stations in the area (my common use for Google Maps), you could see the status of all of your alarms, the pressure of your valves, the temperature of your motors, the status of your locks or just about anything else you could imagine. And then you can zoom down to look at your property or equipment up close.

The interface starts with 3D models. Basically you map the world in 3 dimensional axis and then can map the points on the axis to data. This data can be pixels from an image or data from a any other source. Once mapped, the 3D world can have images superimposed on them to reflect their exact field of view, azimuth, and vertical height. This is done through complex image recognition that searches the 3D world and indexes it to the images in the 2D world, then aligning the two so that they mesh perfectly.

Any camera can be superimposed on the 3D model to let you fly through the world, looking down, across or up from the point of view of the camera. City streets can be flown over showing people walking on the sidewalk, cars driving down the street, or kids playing in their yards. All you need to do is align the camera to the image, map a couple of reference points, and voila, you are done.

The applications for this technology in surveillance are obvious. Central station monitoring companies could manage entire city sectors, tracking criminals from house to house, using the perspective of the cameras they have installed on their customers premise. You could rapidly search the area around an alarm to find where the suspect may be.

In public safety PSAPs and EOC could use this technology to manage traffic, disasters, events, etc.

I thought this was about the best thing ever for surveillance. But last week I saw something even more amazing. Can't talk about it, but it will revolutionize surveillance equally, if not more, than this 3D overlay technology. I only wish I could say more...Suffice to say that if we combine the 3D overlay technology with this other out Earth.

If you want more information on the 3D overlay technology, just let me know and I would be happy to talk to you about it or give a live demo.


  1. let me guess. The ability to track specific video targets... vehicles/people anything that moves and hand off from camera to camera ...coast through non covered areas etc. . Similar to radar target identification and tracking.

  2. Frank this is correct, but it goes much further than that. This technology is ideal for tracking across large geographic areas and watching targets move between cameras seamlessly, as you point out, but it goes much farther.

    The 3D Overlay technology actually maps data in 3 dimensions. The data can be video, or it can be links to access control systems, lighting control systems, traffic control, building automation, sensor networks, or just about anything else.

    So, if I get an alarm at building A and I see the intruder running across the parking lot, I could look ahead of him and close the entry gate, turn on security lights, make an announcement over the address system, or make other changes all from within the interface, where the controls appear where they are physically located. I no longer need to cross reference systems and I can remotely immerse myself in the location I am protecting.