摘要
提出了一种基于包装图像轮廓配准的包装缺陷检测方法。该方法先对包装图像进行轮廓提取,计算目标轮廓的质心,从而得到配准平移参量。围绕目标质心每隔一定角度抽得质心到轮廓的距离值,得到一个距离值的序列。通过计算标准图像和待配准图像的距离值序列的圆周相关的最大值来确定旋转的角度。然后对已配准的包装图像进行缺陷检测,实验证明,本方法对包装缺陷的检测是快速的,有效的。
A new method of detecting package defects based on image contours registration is presented in this paper. In the method, the contour of the object is first abstracted, then calculates the centroid of the object, and obtains translation parameters. Around the centroid, the distances between the centroid and contour are calculated at the same interval of angle, and a series of distances is obtained. By searching the maximum of the circle correlation of the two series of distances about the two package images , the best rotation angle is acquired. Finally, the defects of the package can be detected. The experiment shows this method is fast and effective to detecting package defects.
出处
《包装工程》
CAS
CSCD
北大核心
2007年第3期19-21,共3页
Packaging Engineering
基金
湖南省教育厅资助科学研究项目(05C522)
关键词
包装图像
图像配准
轮廓
圆周相关
package image
image registration
contour
circle correlation
