What's it about?
Homographies are important matrices useful for camera calibration and other applications. This paper shows that a homography can be obtained using only three particular points of the reference plane. We devised a new square-radial checkerboard pattern to detect the required points. The image processing algorithm is based on the Hough transform using parallel coordinates.

Why is it important?
A very simple and robust homography estimation method is proposed. Moreover, we explain relevant concepts such as parallel coordinates and the Hough transform. This work could be useful in other computer vision applications for detection of lines and points of concurrency.

Perspectives
This work offers a staple tool for computer vision and optical metrology researchers. We explain and illustrate useful concepts with the intention of making the paper self-sufficient. The potential of this work can be extended to other applications such as calibration of multiple camera-projector systems and visual navigation.

• See a video sequence processed with this method.
• See a video of quadrilateral detection for a simple augmented reality application.
• Tutorial: PClines Hough transform for line detection (in Spanish).

Full Article More Papers