Extraction and match of visible markers on a planar phantom for fast C-arm calibration

To simplify the flat-panel C-arm calibration in the C-arm based surgery, a fast C-arm calibration method based on automatic extraction of the visible markers on a planar calibration phantom and the markers' match for spatial coordinates is proposed. The method has the steps below: Firstly, the possible markers in images are extracted automatically and their sub-pixel positions are computed by an intensity-weighted centroid algorithm. Secondly, a 2-D affine coordinate system is established in the phantom according to the distribution of the special markers, and false markers are excluded by the distribution pattern. Thirdly, the homography between the image plane and the phantom plane is estimated, and with this homography, the visible markers' spatial coordinates can be inferred directly. Finally, the planar phantom based two-stage method with the radial alignment constrain (RAC) is utilized to calibrate the C-arm. To validate the proposed method, the anti-noise test and the tests for error analysis of markers' re-projection and 3D reconstruction were conducted with the C-arm simulated by a CCD camera and the real C-arm separately. The results showed that the proposed method was accurate, stable and robust, thus its feasibility and effectiveness were verified.