UFO project - please advice

edit

I think that alberto probably hitted the largest problem of your system. Don't forget that you are trying to detect objects in the most varying possible situation. Clouds move, weather type changes, lightning changes. To get a system robust to all this variation will take probably years of fundamental research first. Look for example how many people are in active research for variation invariant detectors nowadays. However, keep up the work if you want to manage this! Just know it will take a long time to get a decent concept of proof :)

Ta guys for your replies. The identification algorithm does not have to be very complicated in the beginning. We can live without it. We will have to look at more videos and say - ah it was a bird. If we have a basic movement detection without identification at the beginning that is good enough to call it proof of concept. From what I saw in mrgloom s reply there a re a few approaches to this.

We can easily exclude planes using http://en.wikipedia.org/wiki/Automatic_dependent_surveillance-broadcast.

About where do I want to deliver this? I want to make this open and as cheaper as possible so ppl all over the world will buy one. I want to give the software free and advice on how to setup the cameras so we can endup with a planetary surveillance system :)

Basically all you need to start your idea is shitloads of cash :) You will need to pay a fulltime developer for writing your code and producing your sets for acquisition.

That does not sound very encouraging. Anyway - this is why I am here basically - to get a quote and a road-map for this project and a time estimate. I would enormously appreciate somebody to look at it as a project not as a job. Somebody who knows how this should be done should be able to say that

• step 1 - create the basic software that detects movement and saves and is build in a modular way with scalability and modularity in mind for future development will take approximately x amount of time and y money.
• step 2 - add a module to do triangulation and get more details about the target will take x1 time and will cost approximately y1 money.
• step 3...

Another option would be to exchange programming work - if somebody is willing to do this kind of work for this UFO project I could make him a website or features to existing websites. Please contact me here or PM if interested in anyway

Well I don't know in the rest of the world, but here in Belgium compagnies ask for a 'feasability study', which is basically a year of cofunded research by innovation centres, which looks if a subject is actually plausible to get. This is a rather cheap way of setting up your small compagny project.

I think that the biggest setback in this project is the fact that there are no UFO's visiting earth

Ahaha - yeah was thinking in the same direction - there are just not enough positive examples XD

But you could the other way - train for known sky object(rockets, planes, birds, meteroids) - and if you have a motion and the prob of that object is below some tresh - then review it. But it could be that an ufo will be be classified as rocket and you still miss it - difficult.

first things .... detect motion ..

absdiff(previousframe, nextframe, n);
absdiff(currentframe, nextframe, l);
bitwise_and(n, l, result);

then you can use "result" as a mask to locate the area on the screen .... then take a deep zoom Area mask and see if the moving object was really a UFO ..... very fun project.