基于向量机的计算机视觉在钢材分类缺陷检测中的应用

Application of Computer Vision on Steel Classification Defect Detecting Based on Vector

随着各领域对钢材工艺水平要求的不断提高,提高视觉检测设备的效率和精度已成为智能制造业的发展趋势和应用方向。基于支持向量机提出一种视觉检测算法,应用于钢材缺陷分类检测中,钢材样本数据集通过最小生成树分类模型处理后,使用最优半径算法选取半径,对稠密及稀疏下的边和点间所存在的关联性加以区分,将获得的分类数据用作支持向量机输入参数。支持向量机模型中选取高斯径向基作为核函数,选取“一对一”分类训练方法,识别钢材流水生产中的各种缺陷。仿真实验证明,支持向量机模型在有新类别出现的情况下,对缺陷类型的分类达到了理想程度,能够满足现有工业领域需求。

With requirements in various fields for the continuous improvement of Steel production process level,improving the efficiency and accuracy of visual detection equipment has become the development trend and application direction of intelligent manufacturing industry.A visual detection algorithm based on support vector machine is proposed,which is applied to steel defect classification detection.After the steel sample data set is processed by the minimum spanning tree classification model,the optimal radius algorithm is used to select the radius,and the correlation between dense and sparse edges and points is distinguished.The obtained classification data is used as the input parameters of support vector machine.In support vector machine model,Gaussian radial basis is selected as kernel function,and"one-to-one"classification training method is selected to identify various defects in steel production.Simulation experiments show that the classification of defect types by support vector machine model reaches an ideal level when new categories appear,and can meet the needs of existing industrial fields.