摘要
精密铸件通过差影法检测缺陷时,图像配准是关键的一步。文中采用一种改进的随机Hough变换和空间数据坐标变换理论相结合的图像配准方法。该方法首先采用OTSU对图像二值分割,利用改进的随机Hough变换检测图像中的多个圆,并把多个圆心作为图像配准的特征点;然后根据特征点对应关系计算配准参数,确立坐标变换关系并对图像进行配准。实验验证了该方法可实现较高精度图像配准。
In the detection of internal defects of precision casting by image subtraction,image registration is a critical step. Based on the modified Hough transform and affine transform in the geometric transformations,the new registration method is proposed. The images are firstly converted to binary image by OTSU. The modified Hough transform is used to detect circles and the center is set to the feature point of image registration. Then the matching parameters is calculated using the corresponding feature points,the transformation relation is created and matches the image. Experimental results indicate that the image matching method has good performance for image matching.
出处
《太原科技大学学报》
2017年第6期424-427,共4页
Journal of Taiyuan University of Science and Technology
基金
山西省回国留学人员科研资助项目(2014-058.)
关键词
图像配准
HOUGH变换
特征提取
仿射变换
image matching, Hough transform, feature extraction, affine transform
