prosaic_alex

prosaic_alex Follow

Computers, old books, and even older books.

9 Followers  21 Follow

Share Share Share

performance test
Here you can see the whole setup in action with a proof-of-concept interface.
After calibration the point was drawn exactly beneath the LEDs, despite changing degrees of distortion in the camera image and beamer projection.
Unreliable detection in this clip is due to the infrared filter missing at the time.
After blobs were detected in the camera picture we needed to find the exact corresponding position in the picture projected back onto the table. This is tricky because the camera and projectors all distort the images. We came up with different methods to undistort them.
The actual camera for LED detection also arrived. For our purpose it needed to detect IR light. To further eliminate obstructions, we placed a filter in front of the lens which only lets IR through.
Here I'm writing the software that would detect blobs of high contrast in an image, testing my own blob properties through a webcam.
After prototyping the main idea for 2 days the actual table plate was built.
While my latest paid project slowly comes to an end I'm feeling like showing some of the milestones we had solving one problem after the other.

Here you can see some infrared LEDs which are filmed by a Kinect camera from above. Software detects the IR light, tracking the position while the LEDs are moved. The observed point is being drawn on screen as the white circle. The screen is being projected from beneath onto the same table the LED object is sitting on.