基于BP网络的红外图像压缩仿真过程分析  

Analysis of Simulation Process of Infrared Image Compression Based on BP Network

作　　者：李颖杰[1,2] 杨华[1] 杨明[1] 王宝荣[1] 张创新[3] 

机构地区：[1]解放军电子工程学院安徽省红外与低温等离子体重点实验室,安徽合肥230037 [2]空军第五飞行学院,甘肃武威733003 [3]解放军电子工程学院501教研室,安徽合肥230037

出　　处：《红外》2009年第6期27-30,共4页Infrared

摘　　要：介绍了BP网络在红外图像压缩中的原理及模型,详细地描述了图像压缩的实现过程,并采用Levenberg-Marquardt算法实现了红外图像的压缩。分析仿真结果表明,通过增加隐含层的神经元数和牺牲压缩率,可以提高重构红外图像的质量。The principle and. model of a Back Propagation （BP） network used in infrared image compression are presented. The implementation process of image compression is described in detail. The compression of infrared images is implemented by using the Levenberg-Marquardt Algorithm. The analysis of the simulation result shows that the quality of the recovered infrared image can be improved when the quantity of neuron in the hidden layer is increased and the compression ratio is sacrificed.

关 键 词：BP网络 LEVENBERG-MARQUARDT算法 红外图像压缩 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]