摘要
钢板在生产及使用过程中产生的表面缺陷不仅影响外观还会降低产品的性能,针对目前检测的效率低、误差大提出了一种结合图像处理与粒子群优化支持向量机的缺陷分类检测系统。利用融合空域的局部二值模式和频域局部相位量化两种特征提取方式的优势对工件的图像进行缺陷特征提取,建立支持向量机(SVM)缺陷分类模型。由于SVM算法参数容易陷入局部最优的问题,所以采用粒子群算法优化SVM的惩罚参数和核函数。在MATLAB 2019b平台进行实验,实验结果对比分析显示,所提算法较传统的SVM分类模型相比提高了18.33%的识别准确率。
The surface defects of steel plate which were made in the process of production and using not only affected the appearance but also reduced the performance of the product.A defect classification detection system combining image processing and particle swarm optimization support vector machine(SVM)was proposed to improve low efficiency and large error.Based on the advantages of local binary mode and frequency domain local phase quantization in feature extraction,a defect classification model of support vector machine was established.Because SVM algorithm parameters were easy to fall into the problem of local optimization,particle swarm optimization was used to optimize SVM′s penalty parameters and kernel functions.The experiment was carried out on the MATLAB 2019 b platform.Comparing the analysis of experimental results,it shows that this algorithm improves the recognition accuracy by 18.33%.
作者
杜绪伟
陈东
Du Xuwei;Chen Dong(School of Mechanical and Electrical Engineering,Qingdao Unive rsity of Science and Technology,Qingdao 266100,China)
出处
《电子测量技术》
2020年第21期122-126,共5页
Electronic Measurement Technology
关键词
表面缺陷
粒子群算法
支持向量机
特征提取
分类模型
surface defect
particle swarm optimization
support vector machine
feature extraction
classification model
