在形态学网络及改进的Ｈａｍｍｉｎｇ网络基础上，提出视觉模式识别网络及训练算法。形态学网络作为特征提取网络，具有相对于图像平移和尺度不变的特性，改进的Ｈａｍｍｉｎｇ网络加速完成特征矢量的分类操作。经过学习的视觉模式识别网络，可完成图像特征的提取，实现相对于图像平移和尺度不变的模式识别。在此基础上，建立了光电视觉模式识别体系，从而实现实时模式识别 Based on morphological network and improved Hamming network, a visual pattern recognition network and training algorithm are proposed. Morphological network as a feature extraction network has the characteristics of being invariant to image translation and scaling, and the improved Hamming network accelerates the classification of feature vectors. After learning the visual pattern recognition network, image features can be extracted to achieve relative to the image translation and scale invariant pattern recognition. On this basis, the establishment of a photoelectric visual pattern recognition system, in order to achieve real-time pattern recognition