基于分块显示的工字型焊件感兴趣区域自适应提取  

Adaptive Extraction of ROI Based on Blocking Display for I Style Weldment

作　　者：石端虎[1,2] 孙修圣[1] 刚铁[2] 张文波[1] 

机构地区：[1]徐州工程学院机电工程学院,江苏徐州221008 [2]哈尔滨工业大学先进焊接与连接国家重点实验室,黑龙江哈尔滨150001

出　　处：《热加工工艺》2013年第5期158-160,163,共4页Hot Working Technology

基　　金：哈尔滨工业大学先进焊接与连接国家重点实验室开放课题(AWPT-M12-05);江苏省高校自然科学基金资助项目(12KJD430003)

摘　　要：感兴趣区域的自适应提取对焊接缺陷评价系统运行速度的提高有重要意义。本文针对工字型焊件射线检测图像的特殊性,提出了一种基于图像分块显示的感兴趣区域自动提取新方法,该方法可精确定位感兴趣区域边缘值,从而实现感兴趣区域的自适应提取。此外,与手动提取方法进行了对比。结果表明:本文提出的方法能很好地实现检测图像中感兴趣区域的自适应提取,且提取结果不受腹板高度变化的影响,算法的鲁棒性强;相比于手动法,提取结果更精确,避免了频繁修改程序的操作,后期图像处理的速度显著提高。该算法的实现使整个图像处理及缺陷评价系统成为了一个完整的整体,工程应用前景广阔。To improve the running speed of evaluation system of weld defects, it is significant to achieve the adaptive extraction of ROI （Regions of Interest）. According to the particularity of X-ray detection images for I style weldments, a new method based on blocking display was put forward, which could extract the ROI adaptively. Using this method the marginal value of ROI could be located exactly, and the adaptive extraction of ROI could be obtained. In addition, the comparison between the provided method in this paper and manual method were carded out. The experimental results show that the adaptive extraction of ROI is achieved perfectly by the provided method. The change of sternum height can not influence the accurate of extraction result, besides the algorithm robustness is high. Comparing with the manual method the extraction result is more precise, and it can avoid the fi＇equent operation of modify program, this can greatly enhance the speed of subsequent image processing. The achievement of the above algorithm makes the whole image processing and defects evaluation system become a full integrality, and its foreground of engineering application is extensive.

关 键 词：工字型焊件 X射线图像 感兴趣区域 分块显示 自适应提取 

分 类 号：TG115.281[金属学及工艺—物理冶金]