基于大数据的低照度微弱点目标激光成像方法  

Laser imaging method of low-illumination weak point target based on big data

作　　者：王华 王玲维 黄汉云[2] WANG Hua;WANG Lingwei;HUANG Hanyun(School of computer science,Guangdong University of Science and Technology,Dongguan Guangdong 523083,China;Hunan University of Technology,College of Mechanical Engineering,Zhuzhou Hunan 412000,China)

机构地区：[1]广东科技学院计算机学院,广东东莞523083 [2]湖南工业大学机械工程学院,湖南株洲412000

出　　处：《激光杂志》2023年第9期182-187,共6页Laser Journal

基　　金：广东省青年基金项目(No.2021A1515110834);广东科技学院创新强校工程项目(No.GKY-2020CQXK-2);广东省高等教育教学改革项目(No.628)。

摘　　要：低照度微弱点目标图像存在亮度低、异常像素点多、噪声严重等问题,为此提出基于大数据的低照度微弱点目标激光成像方法。通过激光扫描得到低照度环境下的图像数据维度,计算三维空间域内的像素点运行轨迹,获取大数据成像信息。采用自适应阈值方法对低照度环境下的成像信息数据进行过滤处理,去除图像高斯噪声,同时保留图像纹理信息数据。将降噪处理后的图像数据输入前级卷积神经网络,完成低照度微弱点目标激光成像。实验结果表明,所提方法熵值保持在1以上,且成像亮度失真度低、对比度高、纹理特征显著,满足低照度环境需求。There are some problems in the image of low illuminance weak point target,such as low brightness,many abnormal pixels and serious noise.Therefore,a laser imaging method of low illuminance weak point target based on big data is proposed.The dimension of image data in low illumination environment is obtained by laser scanning,and the running track of pixel points in three-dimensional space is calculated to obtain big data imaging information.The adaptive threshold method is used to filter the imaging information data in the low illumination environment to re-move the Gaussian noise of the image and retain the texture information data of the image.The image data after noise reduction processing is input to the previous convolutional neural network to complete the laser imaging of low illumina-tion weak point targets.Experimental results show that the entropy value of the proposed method is above 1,and the image brightness distortion is low,the contrast is high,and the texture features are remarkable,which can meet the re-quirements of low illumination environment.

关 键 词：低照度环境 激光线扫描 大数据成像信息 自适应阈值 图像纹理信息数据 前级卷积神经网络 

分 类 号：TN391[电子电信—物理电子学]