基于机器视觉的镍板材表面缺陷检测研究
Nickel Plate Surface Defect Detection Base on Machine Vision
李建华 1刘广鹏 1赵正天 2雷春丽1
作者信息
- 1. 兰州理工大学机电工程学院,甘肃兰州 730050
- 2. 兰州理工大学电气工程与信息工程学院,甘肃兰州 730050
- 折叠
摘要
针对电解法提纯镍板材表面出现的质量缺陷,设计改进Canny的缺陷检测方法.利用双边滤波使图像去噪保边,研究一种增强梯度的掩模并融入Sobel中形成双层卷积核,增强缺陷边缘梯度,弱化缺陷内部及背景区域的梯度.采用分水岭算法代替非极大抑制与形态学边缘连接算法进行边缘细化.实验结果表明:本算法对镍板缺陷的检测效果优于经典算法和其他几种改进算法.
Abstract
To handle the quality defects on the surface of nickel sheet,a defect detection method combining adaptive fractional differentiation and improved Canny algorithm is proposed.Bilateral filtering is used to denoise and preserve the edge of the image.A gradient-enhancing mask is designed and integrated into Sobel to form a double-layer convolution kernel,enhancing defect edges gradient and weakening the gradient inside the defect and in the background area.The watershed algorithm is applied to replace the non-maximum suppression and the morphological edge connection algorithm for edge refinement.The experimental results show that the detection effect of the proposed algorithm on nickel plate defects is better than the classical algorithm and some other improved algorithms.
关键词
机器视觉/表面缺陷检测/改进Canny/最大熵Key words
machine vision/surface defect detection/improved Canny/maximum entropy引用本文复制引用
基金项目
国家重点研发计划项目(2020YFB1713600)
出版年
2024
国家科技期刊平台
NSTL
万方数据