项目名称: 基于仿生视觉感知机理的金属板带表面缺陷在线检测方法研究
项目编号: No.61273170
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 张学武
作者单位: 河海大学
项目金额: 84万元
中文摘要: 金属板带材表面缺陷既影响产品外观,更严重的是损害了产品的抗腐蚀、耐磨、抗疲劳及电磁性能;特别在航空航天、压力容器等高端应用领域,金属板带材表面微小缺陷可能是酿成重大事故的隐患;然而因高速在线检测环境复杂多变,量测信号极易被噪声淹没,缺陷特征弱且难以表征,致使现有视觉在线检测方法检出率低、识别率低。为此,本项目针对高速轧制的金属板带表面缺陷在线精准检测难题,借鉴蝇视觉感知机理,拟建立一套复杂环境下的缺陷在线视觉检测方法:构建仿蝇复眼缺陷在线视觉检测系统,建立对象与传感器的观测模型,设计表面缺陷检测算法,进行缺陷检测性能评估。重点解决(1)仿复眼机制的多传感器配置与任务调度;(2)基于蝇视觉暗盒机制的观测信息的压缩传感和融合;(3)仿蝇视觉通路中高阶神经元信息缺陷特征检测与增强;(4)基于对象特征信息的缺陷辨识方法。为该技术在实际中的应用打下扎实的理论基础和实验验证准备工作。
中文关键词: 机器视觉检测;缺陷诊断;仿生系统建模与信息处理;表面缺陷;
英文摘要: The surface defects on metal strips severely damage the corrosion resistance. Inparticular, the defeats of those applied for aerospace, shipbuilding and pressure vessel, will result in hazard. This project aims to develop an online defect detection approach inspired from a fly visual perception, which are more sensitive to moving target. This approach overcomes the limits of the traditional approaches on defect detection in the complicated environment. Mimicking compound eye mechanism, the following problems are solved: (1) the configuration and scheduling of the multi-sensors; (2) the compressing and fusion of the observed information;(3) the defect detection and enhancement of the high order neurons; (4) the defect identification method based on object characteristic information. The project will provide the supporting principle for the future practical application.
英文关键词: machine vision inspection;defects diagnosis;modeling for bionic system and information process;the surface defects;