基于目标面积特征分析的带钢缺陷图像分割方法  

A steel defect image segmentation method based on target′s area characteristic analysis

作　　者：赵霆[1] 管声启[1] 王鹏[1] 

机构地区：[1]西安工程大学机电工程学院,陕西西安710048

出　　处：《西安工程大学学报》2015年第4期477-481,共5页Journal of Xi’an Polytechnic University

摘　　要：通过分析带钢缺陷图像,提出了一种基于目标面积特征分析的带钢缺陷图像分割方法.首先,将采集到的带钢缺陷图像进行中值滤波处理;然后,通过等分图像灰度范围所确定的一系列阈值对带钢图像进行预分割,计算每次预分割出的目标区域的面积,并找出面积发生突变的区间;在此突变的区间内,按照上述方法对带钢图像缺陷存在的区域进行细分,找出面积突变点,以此确定最佳阈值;最后,通过最佳阈值进行带钢缺陷图像分割.实验证明,将此方法应用于带钢缺陷图像分割过程中,能够完整有效地分割出带钢缺陷区域,为带钢缺陷的视觉在线检测提供了可能性.T hrough the analysis of steel defect image ,a steel image segmentation method based on area characteristic analysis was proposed .Firstly ,pretreats steel defect image was pro‐grammed with median filter algorithm ,and then the steel image was segmented with a series of threshold ,calculating the target′s area and finding the region w here the area shape changes . The region with steel image defects was subdivided with the same method to determine the op‐timal threshold ,with which steel defect image was segmented .Experiments prove that applying this method to steel image online detection can effectively segment the steel defect region and provide a possibility for the online steel vision detection .

关 键 词：带钢图像 图像分割 缺陷检测 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]