NEW ALGORITHM FOR RECOGNIZING CYLINDRICAL PATTERNS BY SURFACE BACKPROJECTION

In this paper, a simple and fast approach is proposed to recognize the unknown cylindrical patterns using backprojection techniques. The method uses a known-height label pasted on the cylindrical object, on which an object pattern (so-called cylindrical patterns) is associated for the robot to identify the desired object. First, a simple camera calibration method is used to compute the camera parameters by a single camera view of the object. A surface backprojection technique is then adopted to reconstruct the pattern on the cylindrical surface of the object. Finally, the reconstructed pattern is matched with those of each object model using the difference of their NRD spectrum. Experimental results to prove the feasibility of the proposed method are also included.