摘要
目的 对X线头影测量分析中标志点实现自动识别。方法 在自动定点过程中,对于不同的标志点采用不同的方法。选择合适的模板匹配方法应用于头影图像,可以确定孤立标志点。对于依附于边缘的标志点采用边缘检测的方法。首先对标准图像的特征边缘利用仿射变换确定被测图像特征边缘的初始位置。然后采用基于仿射变换的模板匹配方法,将模板匹配和仿射变换结合起来,调整图像的初始边缘。最后采用跟踪虫技术对边缘作进一步的修正。根据提取的边缘曲线的特征和标志点的几何特性能够确定依附于边缘的标志点。结果避免了由于头影图像中有些目标在不同图像中的位置存在差异或形变而不能识别目标的缺陷,能更准确地判断目标。结论 该技术能较好的确定X线头影图像中常用的测量标志点。
Objective To automatically Locate landmarks of the X - cephalometric analysis. Method We apply many algorithms due to landmarks that have different characters. Some template matching algorithms can recognize isolated point. Some edge detection algorithms can extract landmarks near boundaries. Firstly, we locate the initial position of boundary by affine transformation that uses four acquired points (four key landmarks). Then we adjust the boundary by template matching based on affine transformation, Finally, we adjust the boundary again by boundary tracking bug which tracks along the acquired boundary. Then we can locate landmarks near boundaries based on the character of boundary curves and landmarks. Result We avoid the error of locating landmarks because of the different positions of some object in different cephalometric images, and more correctlylocate landmarks. Conclusion We locate landmarks of the X - cephalometric analysis very well.
出处
《临床口腔医学杂志》
2002年第5期337-339,共3页
Journal of Clinical Stomatology
基金
江苏省科委社会发展项目BS99063
江苏省333工程2001年
江苏省教育厅留学回国人员2001资助
