基于阈值分割的单板穿孔缺陷识别算法研究

Recognition Algorithm of Single Plate Perforation Defect Based on Threshold Segmentation

单板穿孔缺陷会影响其胶合过程的质量,是单板分选过程中主要的识别目标。基于图像处理技术的图像分割算法将穿孔缺陷区域与背景区域准确地分开是自动分选的重要前提。本研究提出一种基于阈值分割的单板穿孔缺陷识别方法,首先基于彩色图像的RGB空间将采集的图像转换为灰度图并输出R分量灰度图,采用二维中值滤波处理滤除非线性噪声,最后采用阈值分割和连通域处理相结合的方法将背景与目标区域分离。结果表明,遗传算法的最大熵阈值法选取的阈值和试验时间均优于其余3种(迭代阈值法、大津阈值法、最大熵阈值法)算法,结合连通域处理方法可以有效提取单板中的穿孔区域。

The perforation defect of veneer affects the quality of the bonding process and is the main identification target in veneer separation process.An image segmentation algorithm based on image processing technology can accurately separate the perforated defect area from the background area,which is an important prerequisite for automatic sorting.In this paper,a method based on threshold segmentation was proposed to identify the defects of veneer perforation.Firstly,the collected images were converted to grayscale images based on the RGB space of the color images,and the grayscale images of R components were output.Two-dimensional median filter was used to filter the non-linear noise.Finally,a method combining threshold segmentation and connected domain processing was used to separate the background and target region.The experimental results showed that the threshold value and experimental time selected by the maximum entropy threshold method of the genetic algorithm were better than the other three algorithms(iterative threshold method,Otsu threshold method,maximum entropy threshold method)in this study,and the connected domain processing method could effectively extract the perforated region in the veneers.