New real-time labeling and recognition method for multiple infrared markers based on parallel pipeline

A new multiple infrared markers labeling and recognition method implemented by FPGA is proposed. A scanning window and a method to reduce the capacity of equivalent tables handling are proposed according to the characteristics of the connected component shaped by a marker. A series of operation including image smoothing filter, threshold processing, preliminary labeling, second scan and centroid calculation are implemented based on parallel pipelines. Real-time labeling and 2D location recognition for multiple infrared markers are realized in different illuminations. The experimental results show that the latency between the original image data and the processing results is only four pixel clocks, which corresponds to the accuracy of sub-pixel. The proposed system can satisfy the requirements of applications such as surgical navigations that require high time efficiency.