Webcam working, ideas for target?

A lot depends on the environment you're working in. I did some trial runs with three blinking (1hz) blue LEDS in a line. This worked fairly well -- I used frequency and color as the major components to identify rather than the actual pattern.

anyone

An ideal target for a webcam would be a high contrast shape (black on white). Traffic signs are a good example. It might be useful also to have a black border.

It is possible to get a rough idea of distance (size of sign) and directions (distortion of sign).

John Casey

I used blocks of bright colours in a row, about 2cm square. This gave quite a large target to pickup from a distance etc.

The idea was that using 4 different colours I could have 16 different "instructions" or whatever. It is easy to work out the order of the colours just by working out the centre of each colour block along the x axis (i.e. the average position of the pixels it detected with that colour - helps to eliminate the issues of random noise on the camera), and then work out the order of the blocks. This way you only need to detect colour and do some very simple maths to work out where the centres of each block are rather than doing any complex image analysis.

Matt

Any particular kind of motion (in any shape) is easy for a stationery camera to detect; then the computations work out wether it is actually a valid target or not. (Something I worked on a while back).

----------------------------------------------------------------------- Ashley Clarke

It is much easier to get a better idea of distance with stereoscopic vision, comparing the position of the object in one field with its position in the other... Not terribly difficult work, either (assuming you can locate the object reliably in the bitmap).

Thats where all the work is! :-)

Yes, and remember that the video information jumps around alot too (even with a still camera) after it`s been through the digitizer.

----------------------------------------------------------------------- Ashley Clarke