摘要
针对单面平整航空件单件超声检测效率低下问题,采用多试件扫查及多图像平均方法,以提高检测效率。基于多图像平均方法建立了灰度图和二值图像双模板图像。以工件重心为控制点,对二值图像通过变精度最大互相关图像配准算法获得图像配准的旋转角度θ。对灰度图像绕重心旋转角度θ进行配准,再经过减影处理获得缺陷图像,从而实现特征提取和缺陷识别。航空锻件多试件超声检测实验结果表明,该技术可以明显提高航空件超声检测的效率和自动化水平。
In the single ultrasonic test of aircraft components which have one even face, the efficiency is low. To improve test efficiency, multi-specimen ultrasonic test and its average method of multi-images were introduced. First, model image was created through average of multi-images. The model image included one gray-level image and one bilevel image. Bilevel image registration improved the efficiency of image registration. The average of multi-images avoided the disturbance of noise and boundary wave. Then, Alterable precision maximal mutual-information registration algorithm based on barycenter realized bilevel image registration. The rotation angle θ was got. Finally, the gray-level image rotated angle θ around barycenter. Through subtraction processing, the image including flaw information was got. And feature extraction and flaw detect were put in practice. It is shown that the automatization and efficiency of ultrasonic detection can be increased by taking practical measures in the multi-specimen ultrasonic test of forging production for aircraft components.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2005年第11期956-959,共4页
China Mechanical Engineering