一种图像目标实时识别系统  

Real Time Image Target Recognition System

作　　者：刘伏虎[1] 任世宏[1] 朱贵明[1] 

机构地区：[1]北京理工大学电子工程系

出　　处：《北京理工大学学报》1997年第6期717-722,共6页Transactions of Beijing Institute of Technology

摘　　要：以ＴＭＳ３２０Ｃ４０为主处理器，选用不变矩为目标特征，采用以积分方法为基础的一种不变矩快速算法．对极点的计算提出了具体算法．给出了各种极点的叠加分量．并采用ＢＰ神经网络对目标进行分类，实现了一种图像目标实时识别系统．实验表明，识别效果良好．A real time image target recognition system was presented. The main processor of this system was TMS320C40. Moment invariants were used as the features of the target and a fast algorithm based on Integral Method was used to compute the moment invariants. After the feature extraction,a BP neural network was sued to classify different targets.Experiments showed that the system could efficiently recognize the targets.

关 键 词：图像处理 实时目标识别 神经网络 图像分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]