成品家具中的表面死节缺陷图像检测方法  

Image Detection Method for the Dead Knot Defect in the Surface of Finished Furniture

作　　者：程玉柱[1] 李赵春[1] CHENG Yuzhu;LI Zhaochun(College of Mechanieal and Electronic Engineering,Nanjing Forestry University,Nanjing 210037,China)

机构地区：[1]南京林业大学机械电子工程学院,南京210037

出　　处：《家具》2018年第5期106-110,共5页Furniture

基　　金：国家自然科学基金项目(51305207);南京林业大学大学生创新项目(2016NFUSPITP043)

摘　　要：针对成品家具中的表面死节缺陷,提出一种基于局域直方图匹配与水平集的死节缺陷图像检测算法。将RGB彩色图像转换成HSV图像,提取训练样本集,并将样本与分块图像进行直方图匹配,得到匹配相似度灰度图,并用Otsu分割得到初始分割子图。最后对子图采用水平集进行精细分割,得到死节缺陷目标。Matlab试验结果表明,提出的算法能降低光照影响,分割效果好,能很好地提取家具表面死节缺陷。Aiming at the dead knot defects in the finished furniture surface, a surface crack image detection algorithm based on histogram matching and level set was proposed. After the transforming of the color RGB image into HSV image, a training sample set was extracted, and the sample and each block image were subjected to histogram matching to obtain a matching similarity gray scale image, and the initial binary sub-image was obtained by Otsu segmentation. Finally, the sub-images were finely divided by using level sets to obtain dead-knot defect targets. The Matlab experimental results showed that the proposed algorithm can reduce the influence of illumination and has a good segmentation effect and can extract the dead-knot defects in the furniture surface.

关 键 词：成品家具 死节缺陷 直方图匹配 水平集分割 

分 类 号：TS664[轻工技术与工程]