We started with a very simple approach that only required a camera (proliferating on all laptops and desktops) and computer software coded in Visual C++. Since the software was the driving force behind the project, we had to heavily research algorithms since we had no previous knowledge of their functionality. After analyzing over 20 algorithms and trying each one using a brute-force method, we discovered over time that the best way to approach this was to combine multiple algorithms.
It is necessary to have a good camera (high Frames Per Second) for this program to work. The program tracks a contrast in color and uses the tracked contrast as a binary 1 (positive) and everything else as 0 (negative). If the code were to track a shape rather than a color, it may be more effective in helping the camera understand the contours of the hand and it will not delay as much. The results indicate an optimal position, lighting, and color for the current program.
There is a huge impact that the environment has on the program, it won#t work as well with constantly varying background colors and will not work at distances over five feet meaning a background independent code would be preferable. In order for our project to have more potential, the method for tracking must change.
The design and construction of a computer program to aid a user to control and manipulate applications on their personal computer by using the movements of their hands in 3-D space.
Science Fair Project done By Tushar Alladi