Computer Vision Project


Overview
Watching all the goings on in Jackís lab left me with a desire to try building something physical. For my ece432 final project, I built a simple electronic whiteboard out of some consumer electronics that we had lying around the lab. In putting this assignment together Iíve strained my expertise a bit, venturing into a subfield that I know relatively little about Ė HCI. Thus the related work section of my report probably isnít nearly as savy as it ought to be. There are still some interesting places that I can take these algorithms: the application might make a good test case for various GPU particle filter implementations, and it would be interesting to try designing cellular automata (or something similar) that could push the detected images towards thin smooth lines.


The Writeup
Professor Wu requested that we prepare a 30 page report detailing our projects. I have little experience creating such verbose documents, but I think that preparing something that explains the sorts of details that would be left out in an academic paper has been a useful exercise. Here is the pdf (the document included here is not 30 pages, though it would be were it double spaced).


Videos
To complement the figures in the paper, I've made a number of videos showing the system in action.
  • A video showing the use of a laser pointer to draw on a monitor. The setup is very simple, and few precautions have been taken against background interference. In this demo, the first time the calibration algorithms are run, they fail, and need to be repeated.
  • A video showing the data that the webcam sees from behind the back-projection screen. The circular, high intensity red light that shows through the screen allows for easy point tracking, which in turn allows for accurate, low latency interactive drawing.
  • A video showing the calibration of the back projection system.
  • A video showing the back-projection system being used.