论文部分内容阅读
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 sub-pixel linear feature detection method is proposed according to the gradient direction and local characteristics of pixels. Points on a straight line with the same direction in the image have similar gradient directions. Therefore, firstly, they are divided into different point sets according to the gradient direction of the edge points. Then, the pixels forming each straight line are concentratedly scanned at the same point, and then according to the gray distribution of the pixels in the local small neighborhood of the straight line, Finally, the sub-pixel precision fitting point is obtained on the line segment by judging the position relationship between the line segment obtained in the small neighborhood and the line fitting by the edge point, Further fitting to obtain high-precision linear feature location. Experiments show that this method can be more than 0.05 pixel positioning accuracy, angle positioning accuracy over 022 °, and has good real-time.