摘要
该文提出了一种飞机识别的方法,采用轮廓追踪法去除大部分干扰项;用矩不变量的方法描述飞机的特征;对标准Bp网络算法进行改进,使网络训练快速,误差小,从而使整个系统具有识别过程迅速、稳定、准确率高的特点。
This paper provides a kind of method to recognize plane.This method use contour tracking to eliminate much noise.And then,it use moment invariants to describe the feature of the plane.At last,this paper adopts Bp neural networks as its classifier.It has changed the standard Bp neural networks,making it training faster and decreasing residual,to make the whole recognition system recognize faster,more stable and more accurate.
出处
《电脑知识与技术（过刊）》
2009年第5X期3771-3772,3778,共3页
Computer Knowledge and Technology
关键词
轮廓追踪
矩不变量
BP神经网络
contour tracking
moment invariants
BP neural networks
