HomographyLab (Lab is an abbreviation of LABoratory) is a library for Homography Estimation written in C++ combined with OpenCV (both CPU and GPU implementations are available). This library has been evaluated at the TRL 7 (Technology Readiness Level) and is protected by the French APP (Agency for the Protection of Programs). This library implements the homography observers proposed in the papers:
Real-time and robustness (with respect to fast camera motions, occlusions, image blurs, sudden changes in light intensity, poor image quality, etc.) are the two principle distinguished features of HomographyLab and the implemented algorithm with respect to the state-of-the-art codes and algorithms.
The library has been designed with various sub modules written in separate C++ classes like feature extraction and detection, feature matching, nonlinear homography observer. Therefore, the user can also use these individual libraries to write their own custom code and thus save significant time and resources.
This library has been successfully implemented in various robotic and computer vision applications such as image stabilization, station-keeping of an AUV, and linear velocity estimation.
