近红外高光谱成像的建筑材料缺陷自动检测系统  

Building Material Defect Detection System Based on Near Infrared Hyperspectral Imaging

作　　者：周明丽 ZHOU Ming-li(Sichuan Tianyi College,Mianzhu 618200 China)

机构地区：[1]四川天一学院,四川绵竹618200

出　　处：《自动化技术与应用》2024年第7期30-34,共5页Techniques of Automation and Applications

摘　　要：为提升建筑材料缺陷自动检测准确率,提高检测效率,研究近红外高光谱成像的建筑材料缺陷自动检测系统。设计系统硬件,组成包括图像采集传感器、高光谱成像仪及ARM处理器,由硬件实现建筑材料高光谱图像的采集。经近红外高光谱图像黑白校正,降低暗电流、噪音、光源强度引起的干扰后,采用MGS算法提取高光谱图像波长,并将其作为稀疏表示分类器的输入变量,分类器的输出结果即为建筑材料缺陷自动检测结果。实验证明:光谱反射率越高则表明缺陷越严重,可实现建筑材料不同类型缺陷自动检测,检测准确率平均可达96.5%;检测效率不受待检测建材数量影响,检测稳定好。In order to improve the detection accuracy of building material defects and improve the detection efficiency,a building material defect detection system based on near-infrared hyperspectral imaging is studied.The hardware of the system is designed,including image acquisition sensor,hyperspectral imager and ARM processor.The hardware realizes the acquisition of hyperspectral images of building materials.After the black and white correction of the near-infrared hyperspectral image to reduce the interference caused by dark current,noise,and light source intensity,the MGS algorithm is used to extract the wavelength of the hyperspectral image and use it as the input variable of the sparse representation classifier.The output of the classifier is construction material defect detection results.Experiments show that the higher the spectral reflectance is,the more serious the defect is,and the detection of different types of defects in building materials can be realized,and the detection accuracy rate can reach 96.5%on average,the detection efficiency is not affected by the number of modeling materials to be tested,and the detection is stable.

关 键 词：近红外光谱 高光谱成像 建筑材料 缺陷自动检测 目标检测 系统设计 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP181[自动化与计算机技术—计算机科学与技术]