Real-Time Fingertip Detection from Cluttered Background for Vision-based HCI *
Lianwen Jin, Caihong Chen, Li-Xin Zhen, Jiancheng Huang · 2005
Vision-based HCI (Human Computer Interaction) is an important technology to make machine more intelligent. Fingertip tracking and detection is the key for many vision-based HCI systems. In this paper, we propose a new fingertip detection approach from cluttered background, and present a new simple but effective foreground segmentation algorithm based on the characteristics of intensity change over the entire image, and a new background updating method based on foreground connection detection. The fingertip position is finally detected based on template matching after the finger foreground image has been successfully segmented. Based on the accurate detection of fingertip, a novel virtual character recognition system is designed. With this fingertip tracking based system, human can input characters to computer using just the movement of his fingertip, without any additional devices such as keyboard, digital pen, so that it provides a novel wireless character inputting modality for HCI