球笼内表面裂纹自动识别系统图像预处理研究

Image preprocessing of CV surface cracks automatic identification system

介绍球笼内表面裂纹荧光磁粉自动识别系统的工作原理和结构组成,对系统紫外线光源波长进行分析和研究,提出该系统所用的紫外线光源波长范围应在320nm^390nm之间。针对球笼的特殊构造,重点介绍了底面和侧面分离的球笼内表面图像采集方式。通过对CCD采集的图像进行7×7模板的中值滤波,Sobel边缘提取,阈值二值化等一系列图像预处理,并计算预处理后图像中缺陷的特征值,可以有效地识别出球笼的裂纹。

This paper introduced the structure and the principle of the fluorescent magnetic CV surface defects automatic identification system and analysed the UV light in the system. Due to the CV special structure,the improved method of the CV inner surface acquisition by CCD was focused on. This system used side and bottom separated method to take image. A series of image preprocessing was applied in the CCD captured image,such as 7 × 7 templates median filter,Sobel edge detection,threshold binary,and it calculated the eigenvalues of image defects. Finally,the system can effectively identify the cracks cage.