Software as Sensor

I am seriously intrigued by the notion of using software as the key enhancement to already existing (and hopefully simple) sensors.  Think already existing video cameras being used in ways that hugely extend their usefulness simply through the use of algorithms to process the image data.  I know I'm a bit late to the party here but, even so, wanted to capture my thoughts.

This line of thinking was driven by seeing a NY Times article (which I can't find now), but perhaps even better than that link is a link to the MIT Website.  Pretty fascinating stuff.  How else could this be extended?  Let's take things One Step Beyond . . . 

Simple extensions (tiny steps):

• Pulse monitoring on Treadmills and Exercise Bikes:  As long as your face is visible to a presumably cheap camera, your pulse should be available.
• Threat Monitoring: TSA, stop scanning us by shooting all sorts of wonderful radiation and start videotaping movement and faces and bodies.  Surely there must be some tell-tale signs of having a gun stuffed in your pants that would be obvious when you are magnifying movement by 100x?  Perhaps there should be separate face, torso and leg video scanners each looking independently for threats.  Perhaps this wouldn't be sufficient, but cheaply additive to existing measures.  Would probably work best if you make people walk at a normal stride for several paces prior to getting to the metal detector.  It seems as if the TSA could easily add some barriers/ropes and a stop line to make this happen.
• Pilot/Driver Monitoring: I think some of these systems are in place already on high-end cars, but this could potentially make it much less expensive to roll-out to our commercial trucking, bus and aviation industries.  Especially when large numbers of people are involved, the vitals of our pilots ought to be monitored.  Eventually (with much experience), roll this out to all cars. Add distracted driver alerts.  These alerts could be triggered not only by texting or email, but also by overly engrossing conversations. 

More complex extensions:

• Body fat measurement:  Maybe I'm obsessed with this one, but it seems to me that all of the measurement technics are pretty terrible.  Either crazy expensive or not super accurate nor descriptive of distribution of fat.  Maybe video isn't necessary here, but still pictures would work.  First thought is to include age, height and weight and then mix some measurements of body parts (including "hard measurements" - places with likely little fat buffering - to get a clean view of the skeletal frame) with pictures of the individual from various angles (i.e., front on and view from every 45 degree increment).  Video would interesting here because presumably you'd get the view from many more angles.  An even further refinement would be to add pictures of the individual with a series of rubber bands on.  These would be applying a known amount of force (pressure on the skin) and thus able to measure the surface squishiness.  Ideally, this results not only in a reasonably accurate aggregate assessment of the individual, but also a view as to the distribution of fat in their body.  Fat mixed among one's internal organs would obviously not be measured easily using this method.  But given that its primarily software driven, I bet a package could be put together (with a digital tape measure) for only $30 or so.  Bonus step: instead of taking stills, take a video and have the digital tape measure slowly retract or tighten (similar to a blood pressure monitor) and calculate the squishiness based on the air pressure.
• 3D Modeling of Objects: Spin an object on a spindle, strip out the spindle and go.  3D scanners already exist and will be a critical adjunct to 3D printing.  Some software scanners also already exist.  Reasonably useful, too, from my exploration.  That said, get your 3D scanner software fast, before some knucklehead tries to prohibit via DMCA or some such nonsense.
• 3D Modeling of buildings: Similar to above, but of "inside" surfaces instead of outside surfaces.  How cool would it be if you could walk through your house with a video camera and make a complete 3D model (think, you are trying to sell your house).  This will happen via consumer grade (think, mobile phone) video soon and at much lower cost than is happening now with complex laser scanners (of crime scenes, for instance).
• Consumer Speed "Trap": Tired of your neighbors driving too quickly down your road?  Set up a camera in your front yard and then process both the license plates and the speed of the car (easily detectible from identification of the make and model - and its interaction with the optics of the particular video camera) by comparing successive frames.  Upload to one of the "bad driver websites".

More to come (maybe).