Homography estimation by two PClines Hough transforms and a square-radial checkerboard pattern

The homography matrix is crucial for many optical metrology applications, such as three-dimensional surface imaging by structured-light projection. In this paper, a straightforward homography estimation method using the image of three particular points on the reference plane is proposed. For this, a new square-radial checkerboard pattern designed to generate three sets of concurrent lines is proposed. The lines and points of interest are detected using two PClines Hough transforms. Relevant concepts such as parallel coordinates and the used Hough transforms are explained. The usefulness of our proposal is verified experimentally.