FFT-Based Orientation Recognition Algorithm in Mirosot

The accuracy of the object recognition is the basis for the decision-making subsystem in the centralized-control Mirosot system. This paper proposes a novel cross-power spectrum based approach to detect the orientation of the robots more accurately. In the proposed algorithm, the images of all objects are firstly gotten by image segmentation technologies from the current-frame image. Then, the object edge image and its referenced edge image are transformed into log-polar coordinate space after FFT, by which rotation and scaling in Cartesian coordinate space can be reduced to 2-D translation in log-polar coordinate space. Finally, the orientation of the robot can be estimated by phase correlation technique. The experimental results show that the proposed method is of high accuracy with the absolute error less than 2⁰ as well as of good robustness.