论文部分内容阅读
根据智能装配系统的实际要求,提出了一种利用神经视觉进行三维物体识别的理论和方法,在利用立体象对重建物体的三维外形的基础上,建立物体的区域图,利用物体的三维矩及其不变性来构造代表物体的特征矢量.采用ART2神经网络构成神经网络分类器,把物体的特征矢量作为神经网络分类器的输入,从而对物体进行识别或分类.这种识别或分类方法可以在线学习,能满足智能装配环境下连续作业的要求
According to the actual requirements of intelligent assembly system, a theory and method of using neural vision to recognize three-dimensional objects is proposed. Based on the use of stereo pair to reconstruct the three-dimensional shape of objects, an area map of objects is established. Its invariance to construct the feature vector representing the object. The ART2 neural network is used to form the neural network classifier, and the feature vector of the object is taken as the input of the neural network classifier to identify or classify the object. This method of identification or classification can be learned online to meet the requirements of continuous operations in an intelligent assembly environment