摘要
在对铝合金铸件进行X射线实时探伤时,由于采集回来的许多图像的目标物体的轮廓比较模糊,这样在评片时容易造成对铸件的误判和漏判。针对这种现象并结合这类图像噪声源多的特点,在研究常用卷积算法的基础上,采用特殊卷积(特殊的卷积核)操作,同时根据雕刻的艺术底色为灰色,最后给卷积结果加上灰度值128,实验结果表明图像中目标物体的轮廓变得更清楚,同时处理后的图像具有比较好的三维效果,这给工作人员在看图时带来了方便。
When carrying on X- ray real- time crack detection to the aluminium- alloy cast, because a lot of goal object in collected images has blurring contour, it is apt to cause the erroneous judgement to the cast and omission to judge while commenting one like this. Aiming at this kind of phenomenon, combining the characters of many image noise fountainheads, on the basis of studing general convolution, using the special convolution (special convolute kernel) to operate, at the same time because of gray background colour of carving art, at last all convolution results are added in 128 grey- value. The experimental result indicates that the contour of the goal object becomes clearer in the picture, at the same time disposed images have better three - dimensional effect. It' s more convenient for staff to see them.
出处
《信息技术》
2006年第11期109-111,共3页
Information Technology
关键词
雕刻
铝合金铸件
卷积
轮廓
carving
aluminium alloy cast
convolution
contour