A novel edge detection approach used for online dimensional measurement of heavy forging
A novel edge detection approach used for online dimensional measurement of heavy forging
复制标题
DOI：
10.1117/12.807046
复制
发表时间：
2008-12
期刊：
影响因子：
--
通讯作者：
Chunhai Hu;Xiaoxue Song
中科院分区：
文献类型：
--
作者：
Chunhai Hu;Xiaoxue Song
提出了一种能够客观检测高温锻件图像中显著边缘的边缘检测方法。观察高温物体时可能出现的图像退化问题令人严重关切。本文主要研究了立体视觉技术在在线尺寸测量中的应用。介绍了高温锻造的特点。采用全局自适应阈值预处理消除大部分冗余的鳞片区域，从复杂背景中分割出目标。然后，特征边缘检测和定位在单像素尺度。作为最后的附加步骤，应用环绕抑制后处理，通过从边缘图中提取真正的特征边缘来提高边缘检测性能。整个方法在真实的热锻图像上进行了实验，实验结果证明了该方法的有效性。该方法是专门为大型锻件的在线尺寸测量而设计的，但一般足以应用于任何高温物体的其他边缘检测。
An edge detection method was developed with capability of objectively detecting significant edges in images of high temperature forging. The issue of potential image degradation when viewing hot objects was serious concerned. The paper was absorbed in online dimensional measurement using stereovision technology. Particular characteristics of high temperature forging were described. A global self-adaptive thresholding preprocessing was used for eliminating most redundant mill scale regions and segmenting object from complex background. Then, the feature edges was detected and localized at single pixel scale. Post processing of surround suppression as final additional step was applied to improve edges detection performance by extracting genuine feature edges from edges map. The entire method was performed on a real hot forging image and the performance had been proved in experiment result. The approach was specifically designed for using in online dimensional measurement of heavy forging, but generally enough to be applied to other edge detection of any high-temperature object.