神经网络的二值图像模式识别技术  

Two value image pattern recognizing technology of ANN

作　　者：洪流[1] 张荫锡[1] 

机构地区：[1]中国空空导弹研究院,河南洛阳471009

出　　处：《红外与激光工程》2001年第6期432-437,共6页Infrared and Laser Engineering

摘　　要：文中论述的是以攻击空中目标为背景的近场图像探测装置对目标的一次性探测机理和识别原则 ,并根据系统的性能指标要求 ,在图像生成过程中对目标的局部图像进行识别和处理。首先介绍了运动目标的一次性非完整图像的基本特征 ,图像的像点连接键描述和矩阵表示及不变性处理原则。再根据其识别准则 ,论述了一次性非完整目标图像基于人工神经网络的二值图像模式识别方法 ,这样提高了近场图像探测装置对目标的自动识别和抗干扰能力。文中还给出了人工神经网络二值图像模式识别处理系统的原理框图和仿真结果。In this paper, the one-shot recognition principles and detecting principles of the near field image detecting of aerial targets are illustrated. Based on the capability requirments of the system, the partial image of the targets must be identified and processed simultaneously during the generation of the image. Firstly, the primary characteristics of moving targets' one shot partial images, pixels link bonds, matrix expression and the invariable principle are illustrated. Then, based on the recognition rule, the two value image pattern identifying method of one shot partial target's image on the basis of ANN is introduced. So the automatic identifying and anti-jamming capability of near field image detecting device is highly improved. Furthermore, the simulating results and the block diagrams of ANN two value image pattern recognition system are presented. \;

关 键 词：二值图像 人工神经网络 模式识别 图像识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]