Abstract: This paper describes a full-automatic pattern recognizing and tracing system based on neural network. After picking up their vectors by the MOD arithmetic, it classifies the patterns of the mobile objects according to neural network theory. The paper meanwhile gives the experimental result of auto recognizing and tracing process for multi-objects moving under complicated background.

Key words: Neural network; Moving object detector; Picking up characters; SOFM; B-P net

摘要： 建立了一个基于神经网络的全自动模式识别跟踪系统。通过MOD 算法提取出各个运动目标的特征向量后，利用神经网络对各个目标自动进行模式分类。文中同时给出了在复杂背景下多个运动目标自动跟踪与识别的实验结果。

关键词: 神经网络；MOD；特征提取；自组织特征映射；B-P 网络