摘要
针对低对比度的金属零件弱边缘提取缺失和缺陷漏检等问题,提出了基于分段幂次Retinex和含惩罚项自适应鸽群优化的弱边缘提取和缺陷检测两阶段算法。一阶段使用分段幂次Retinex(PPR)提高弱边缘对比度,构建低通自适应双边滤波抑制和去除无关噪声,使用含惩罚项的自适应鸽群优化分割算法提取边缘;二阶段在提取出的边缘区域内使用不含PPR图像增强的滤波和分割等算法来检测划痕和凹洞缺陷。试验结果表明,两阶段算法在增强保护低对比度弱边缘分界信息后,可完整分割提取出零件的轮廓边缘,进而快速检测出不同位置的零件缺陷。
Aiming at the problems of missing weak edge extraction and missing defect detection of low contrast metal parts,a two-stage algorithm for weak edge extraction and defect detection based on piecewise power retinex and adaptive pigeon-inspired optimization with penalty method is proposed.In the first stage,piecewise power retinex(PPR)is used to improve the weak edge contrast,a low-pass adaptive bilateral filter is constructed to suppress and remove irrelevant noise,and an adaptive pigeon-inspired optimization segmentation algorithm with penalty method is used to extract the edge.In the second stage,algorithms such as filtering and segmentation without PPR image enhancement are used to detect scratches and pits in the extracted edge region.The experimental results show that the two-stage algorithm can completely segment and extract the contour edges after enhancing and protecting the boundary information of low contrast and weak edge,and then can quickly detect the defects at different positions.
作者
陈东
马兆昆
张新伟
Chen Dong;Ma Zhaokun;Zhang Xinwei
出处
《工具技术》
北大核心
2022年第9期128-133,共6页
Tool Engineering
基金
山东省重点研发计划项目(218GNC112007)。
关键词
边缘提取
分段幂次Retinex
低通自适应双边滤波
惩罚鸽群优化
edge extraction
piecewise power retinex
low-pass adaptive bilateral filtering
penalty pigeon-inspired optimization
