Sub-pixel X-marker detection by Hough transform
Ali Yazdani, Hojjat Aalizadeh, Farshid Karimi, Saeed Solouki, Hamid Soltanian‐Zadeh · 2018
High precision center detection of X-markers is required in many applications such as navigation surgery systems and camera calibration. Hough transform is a preferable tool for extracting intersecting lines in an image, which leads to center detection. In this paper, we detect X-marker centers by the sub-pixel precision, using Hough transform. Switching to Hough space helps us to apply processes like thresholding, filtering and weighted averaging on coordinates. The algorithm involves two parameters `Hough Size' and `Filter Size' required to be adjusted for best performance of the algorithm. A dataset of 900 images is used and best performance is achieved by values of 180 and 23 for the above parameters, respectively. Using this setting, 90.8% of the centers are detected successfully by the sub-pixel precision. The average distance between detected centers and reference centers is 0.51 pixels. This suggests that the proposed algorithm has the potential to be utilized for sub-pixel marker detection.