Multiple infrared markers based real-time stereo vision positioning system for surgical navigation

This paper studies the development of a real-time stereo vision system to position multiple infrared markers attached to a surgical instrument. A parallel pipeline algorithm in FPGA is developed to recognize the markers in both image planes and give each of them a unique label; the pipeline architecture includes smoothing filter, thresholding, connected component labeling, and centroid extraction. A distortion correction and stereo triangulation method is proposed to position the markers in space, which is carried out by a cross computation mechanism in a dual-core DSP. The proposed system can recognize the 3D coordinates of four infrared markers with a delay of 8.67 ms. Furthermore, it is capable of positioning 80 infrared markers at most without frame dropping. The performance of the proposed system can meet the requirements of applications such as surgical navigation which needs high real time capability.