Vision-based finger detection, tracking, and event identification techniques for multi-touch sensing and display...

Area of Science:

• Computer Vision
• Human-Computer Interaction
• Image Processing

Background:

• Multi-touch sensing is crucial for modern interactive displays.
• Existing systems often struggle with varying ambient light and noise.
• Robust finger detection and event identification are needed for enhanced user experience.

Purpose of the Study:

• To develop an efficient, low-cost vision-based system for multi-touch sensing.
• To enable accurate finger detection, tracking, and event identification.
• To create a system adaptable to diverse environmental conditions.

Main Methods:

• Utilized bright-blob segmentation with automatic multilevel histogram thresholding for touch point extraction.
• Employed connected-component analysis to identify touch blobs.
• Implemented a two-phase blob tracking and event recognition process analyzing spatial-temporal features.

Main Results:

• Demonstrated robustness against varying ambient lighting and infrared noise.
• Successfully detected and tracked finger movements, including gestures like pressing, hovering, and clicking.
• Achieved accurate identification of touch events through spatial-temporal analysis.

Conclusions:

• The proposed vision-based system is feasible and effective for multi-touch applications.
• The system offers adaptability and reliability in various operational environments.
• This technology enhances the performance and user interaction of multi-touch displays.