| 摘要: |
| 提出一种基于形态学理论的红外热像分割方法,用于工件表面缺陷的自动检测。首先在含缺陷钢制试件红外 热成像检测试验的基础上，对I:件的红外热像进行灰度化、卨斯高通滤波、对数变换和二值化等方法相结合的增强 处理；然后采用形态学方法，基T缺陷的空间连续性和缺陷与噪卢的尺寸差别，设定连通分量所含像素数的阈值，最 终实现红外热像的有效分割。结果表明,新的红外热像处理方法n]?以实现缺陷位置和形状的精确检测，可作为含缺 陷部件的红外检测和自动识別手段。 |
| 关键词: 形态学 红外图像 缺陷检测 连通分量 图像分割 |
| DOI：10.3969/j.issn.1673-5005.2012.03.025 |
| 分类号:TN 219 |
| 基金项目:国家自然科学基金项目(50679083 ) |
|
| Infrared thermal images detecting surface defect of steel specimen based on morphological algorithm |
|
XIE Jing1,YANG Xiao-yan2,XU Chang-hang1,CHEN Guo-ming1,GE Su-an3
|
|
(1.College of Electromechanical Engineering in China University of Petroleum, Qingdao 266580, China;2. Qingdao China Petroleum EDI Safety & Environment Protection Company Limited, Qingdao 266071, China;3. Northwest Oilfield Energy Saving Monitor Center of CNPC，Karamay 834000，China )
|
| Abstract: |
| An infrared thermal image processing framework to detect surface defect of a steel specimen was proposed. It includes two steps: First, gray processing, Gaussian high pass filter, logarithmic transformation and thresholding were used on the original infrared thermal image in sequence for contrast enhancement ； Second, based on the spatial continuity of defect and the size difference between noise and defect, a segmentation method l>ased on morphological algorithm was applied. The threshold number of pixels contained in connected component was set and the effective segmentation of the infrared thermal image was realized. Experimental results show that the proposed framework has very promising segmentation performance and can obtain precise defect information of a steel specimen. It can be used as infrared detection and automatic identification means for components with surface defects. |
| Key words: morphological algorithm infrared image defect detection connected component image segmentation |
