Graphical fiducial marker identification for augmented reality, virtual reality and robotics

摘要

Systems and methods for fiducial marker identification are provided. In the fiducial marker identification system, process 132 is performed in a computing device's GPU 122 and CPU 124 without limitation. To increase efficiency, processing blocks of the more computationally expensive processing 132 are implemented in GPU 122. Additionally, by implementing at least some of the processing blocks in GPU 122, two or more processing blocks are implemented in parallel. The process 132 itself, if implemented on the CPU 124, includes processing blocks that reduce computation time compared to conventional approaches. [Selection diagram] Fig. 1B