散射介质后物体分类方法研究  

Research on Classification of Objects after Scattering Medium

作　　者：孙蕴晗 王永成[1] 徐东东[1] SUN Yunhan;WANG Yongcheng;XU Dongdong(Changchun Institue of Optics,Fine Mechanics and Physics,Chinese Academy of Science,Changchun 130033)

机构地区：[1]中国科学院长春光学精密机械与物理研究所,长春130033

出　　处：《长春理工大学学报（自然科学版）》2023年第3期1-7,共7页Journal of Changchun University of Science and Technology(Natural Science Edition)

基　　金：国家重大科研项目(902017000590)。

摘　　要：实际生活中,观测物体的过程会受到散射的影响,在一定程度上会干扰对目标信息的获取和判断。为了解决散射介质后的目标分类方法适应性差及传统特征提取和分类算法效率低的问题,提出利用VGG16网络对经过散斑自相关处理后的数据进行分类识别的方法。首先从理论上分析自相关方法对散斑图像处理的可行性,再通过计算机仿真实验获取数据集并对其进行自相关处理,最终利用VGG16网络进行训练和测试,测试准确率为95.5%。In real life,the process of observing objects will be affected by scattering,which will interfere the acquisition and judgment of target information to a certain extent.In this paper,the existing techniques for obtaining target information from disordered scattered light field are described and analyzed.The principle of speckle formation and the concepts of optical memory effect and speckle autocorrelation are introduced.In order to solve the problems of poor adaptability of target classification method after scattering medium and low efficiency of traditional feature extraction and classification algorithm,a method of classification and recognition of data after speckle autocorrelation processing using VGG16 network was proposed.Firstly,the feasibility of the autocorrelation method for speckle image processing was analyzed theoretically.Then,the data set was obtained by computer simulation experiment and the autocorrelation was processed.Finally,the VGG16 network was used for training and testing,and the final test accuracy was 95.5%.

关 键 词：散射 光学记忆效应 散斑自相关 神经网络 

分 类 号：TH744[机械工程—光学工程]