In shape identification of the 3-D object using a vision sensor, it is important that the system is insensitive to the translation and the rotation of the object. The authors have developed a system for quickly identifying the shape of the 3-D object, using a dot-matrix-like bright spots projection. In the system, the range data is classified, using the 3-D information of the bright spot. The polar coordinates transform is performed in the frequency domain of the classification data. A pattern of invariance to the translation and the rotation of the classification data is produced, using the power spectrum of the transformation data. The pattern is identified by a multi-layered neural network.
This paper describes the method of the shape identification of the 3-D object, using the dotmatrix-like bright spots. The results of the experiments, which were performed to verify the effectiveness of the proposed method, are discussed.
抄録全体を表示