摘要
角点包含丰富的图像几何信息,其检测精度直接影响后续图像参数的计算与分析。为满足工程上要求的高精度计算出零件轮廓参数,在检测出零件轮廓边缘后,角点定位成为影响结果精度的最重要因素。为此,本文提出了一种基于亚像素思想,综合图像灰度信息与边缘轮廓点特征的高精度角点检测法。该方法使用Marr边缘检测算子得到连续单像素边缘,结合亚像素思想对各个边缘点进行精确定位,在此基础上根据角点曲率最大原理进行检测,最终确定独立角点。实践证明,该方法能有效、高精度地进行角点检测。
Corner contains rich geometric information of an image. So comers play important roles in post-calculation for other geometric parameters. After the part edge was accurately detected, comer location is the most important part for us to calculate some high -precision contour parameters. For this reason, we propose a based -on subpixel approach, which integrates the gray level of the image with the feature of the edge contour. Firstly, use Marr operator to get the continuous single-pixel edge. Then, further locate the edge based -on subpixel method. Lastly, gain independent comers according to the maximum curvature principle.
出处
《仪器仪表用户》
2005年第2期81-83,共3页
Instrumentation
