基于Matlab神经网络的水下目标识别  

Underwater Objects Identification Based on Neural Network in Matlab

作　　者：胡红波[1] 邱继进[1] 马爱民[1] 

机构地区：[1]海军大连舰艇学院,辽宁大连116018

出　　处：《情报指挥控制系统与仿真技术》2005年第5期52-54,共3页Information Command Control System and Simulation Technology

摘　　要：为了实现对水下目标的识别,在现有特征提取方法的基础上,提出了目标像素灰度分布特征提取方法,并针对水下光信号衰减的情况,采用了色彩“补偿”的解决方案。对图像矩阵通过奇异值分解得到了目标有效维特征向量,分类器设计采用了Matlab环境下的BP神经网络,识别结果是令人满意的。这对猎雷具装备的发展具有一定的参考价值。In order to identify underwater object, the paper puts forward a method of getting eigenvalue with pixel grey distributing of a real-time visual image in despite of having ready-made means to get eigenvalue and solves the problem of optical signal attenuation by color compensation, then gets characteristic vector by singular value decomposition （SVD） of image matrix and designs classifier with BP neural network in Matlab. The result is acceptable when identifying. It is of reference value for the development of mine hunting gear.

关 键 词：目标识别 灰度分布 奇异值分解 BP网络 

分 类 号：O235[理学—运筹学与控制论]