mkalten/reacTIVision
computer vision framework for tangible interactive surfaces
What it solves
reacTIVision is a computer vision framework designed for the fast and robust tracking of fiducial markers (physical labels) and multi-touch finger tracking. It specifically targets the development of table-based tangible user interfaces (TUI) and multi-touch interactive surfaces, such as the reactable modular synthesizer.
How it works
The application uses a camera to capture real-time video and identifies the ID, position, and rotation of fiducial markers, as well as round white regions interpreted as fingers. It also supports "blob tracking" for untagged objects. The system transmits this tracking data via the TUIO protocol (using UDP, TCP, Websocket, or Flash Local Connection) to client applications. To handle wide-angle lens distortion, it includes a calibration tool to correct the image alignment and distortion.
Who it’s for
Developers creating interactive installations, tangible user interfaces, and multi-touch surfaces who need a cross-platform tool to convert physical object movements into digital data.
Highlights
- Multi-modal tracking: Supports fiducial markers (Amoeba and Yamaarashi sets), finger tracking, and general blob tracking.
- TUIO Protocol: Implements the TUIO protocol for standardized communication with client applications across various languages (C++, C#, Java, Processing, etc.).
- Lens Correction: Built-in tools for correcting wide-angle lens distortion and image alignment.
- Cross-platform: Runs on Windows, Mac OS X, and Linux with support for a wide range of camera drivers (WDM, UVC, Video4Linux2).
- C-based performance: High-performance real-time processing with configurable image equalization and thresholding.