数字图像直线特征的亚像素位置检测  

Subpixle location detecting of line feature in digital image

作　　者：于新瑞[1] 王石刚[1] 王高中[1] 邵萌[1] 

机构地区：[1]上海交通大学机械与动力工程学院,上海200030

出　　处：《光学技术》2004年第2期138-141,共4页Optical Technique

摘　　要：根据像素点的梯度方向和直线局部特性提出了一种新的亚像素直线特征检测方法。图像中具有相同方向的直线上的点具有相近的梯度方向。因此首先根据边缘点的梯度方向把它们划分为不同的点集;然后在同一点集中扫描得到组成每一直线各自的像素点;再然后根据直线上局部小邻域内像素灰度分布特性,在直线上边缘点的3×3邻域内得到其中的线段解析表达式;最后通过判断在小邻域内所得线段和由边缘点拟合得到直线的位置关系,在线段上得到亚像素精度的拟合点,进一步拟合得到高精度的直线特征位置。实验表明,这种方法的定位精度可以超过0.05像素,角度定位精度超过0 02°,并且具有较好的实时性。A new method is proposed according to the pixel's gradient direction and the line's local property. Since points on lines that have same angles have close gradient directions, the edge points are firstly divided into different collections according to their gradient directions. Then all points belonging to the same line can be scanned in the collection. Third, according to the local straight property in the 3×3 neighborhood of the edge point on the line, analytic expression of the line segment is gotten in it. Finally, the subpixel fitting points are gotten on the line segment according to the positions of the line segment and the line that is fitted by the edge points. Therefore precise location of the line feature is fitted by the fitting points. Experiment results show the method locate the positioning accuracy is better than 0.05 pixel and 0.02 degree, and the operating speed is fast.

关 键 词：哈夫变换 梯度方向 直线特征 图像谊测 亚像索 数字图像 

分 类 号：TN911.7[电子电信—通信与信息系统] TP391.41[电子电信—信息与通信工程]