论文部分内容阅读
快速准确地获取导航信息是实现机器人实时跟踪的关键。为能够有效解决实时参考轨线提取问题,根据轨线的宽度在图像中具有均匀一致的特性,提出了一种快速的参考轨线检测算法。先将图像做二值化处理,然后对图像的每一行在水平方向上进行边缘检测,若相邻的2个边缘点之间的距离满足轨线宽度减2,并小于轨线宽度加3,则这2点就可能是轨线的边界点,同时计算出它们的中点。对图像的若干行进处理得到这些中点后,结合基元法,进行列基元检测,获得基元中心坐标,再对这些做快速Houg变换,得到轨线的直线方程在参数空间的映射参数。针对实际轨线图像检测实验表明,该算法能快速鲁棒地检测出参考轨线。
Acquiring navigation information quickly and accurately is the key to real-time tracking robot. In order to effectively solve the problem of real-time reference trajectory extraction, a fast reference trajectory detection algorithm is proposed based on the uniform width of trajectory in the image. The image is first binarized, and then the edge of each line of the image is detected in the horizontal direction. If the distance between two adjacent edge points satisfies the track width minus 2 and is less than the track width plus 3, Then these 2 points may be the boundary points of the trajectory, and calculate their midpoints at the same time. After some of these images are processed, we get the center coordinates of the elements by combining the primitive method and the Houg transformation of these elements, and get the mapping parameters of the linear equation of the trajectory in the parameter space. Experiments on actual trajectory images show that this algorithm can quickly and robustly detect reference trajectories.