PVC建材表面缺陷检测系统研究与设计

为满足PVC建材表面缺陷在线系统高速、高分辨率的检测要求,提出PVC建材表面缺陷检测的系统设计方案。为保证检测系统的实时性,硬件上采用面阵CCD摄像机与LED面光源相结合的图像采集装置,软件上采用实时处理和准实时处理相结合的图像处理方法。针对缺陷图像高噪声、直方图单峰明显的特征,提出边缘检测分割技术与数学形态处理相结合的算法流程。通过分析图像缺陷的统计特性,提取PVC建材表面缺陷区域的特征参数,利用基于决策树的分类器将缺陷分为气泡、色痕、裂痕等。实验结果验证算法的有效性和实时性,其缺陷的检出率达到95%以上,识别率达90%以上。

Research and Design on Surface Defect Detection System for PVC Building Materials

In order to satisfy the requirements in high speed、high resolution for PVC building materials line,this paper projects a system designing solution for PVC building materials surface defect inspection.In order to ensure the real-time in detection system,the capture image devices combining with area-array CCD camera and LED surface light source are used in the hardware,the image processing methods combining with real-time processing and quasi-real-time processing are used in the software.For the characteristics of high-noise on defect image and exactly single-peak histogram,Arithmetic flow combining with edge detection segmentation and mathematical morphology processing is presented.By analyzing the statistical properties of the image defects,feature parameters in the area of PVC building materials surface inspection are extracted,the method classifies defects into bubbles、color marks、cracks by using classifier based on the decision tree.The experimental results show the effectiveness and real-time of the algorithm,the detection rate of defects is higher than 95% while the classification rate of defects is higher than 90%.