摘要
针对四旋翼飞行器在室内环境下识别地面线状结构的问题,提出了一种改进的基于单目摄像头,提取线段并跟踪的方法;首先在图像中设置感兴趣的区域(Region Of Interest,ROI),基于HSV颜色空间,根据颜色阈值提取出线段颜色;然后对ROI进行高斯滤波处理过滤图像的噪声,通过Canny边缘检测算法提取出线段的边缘;最后用概率霍夫变换(Probabilistic Hough Transform,PHT)从预处理图像中拟合出直线,计算直线的方向向量实现四旋翼的路径跟踪;实验结果表明:该方法有效地提高了路径识别的效率,使四旋翼能够精确跟踪室内路径。
Aiming at the problem of detecting linear shaped structures in a landscape for indoor quadrotor aircraft, an improved method based on monocular camera is provided for line detection and following. Firstly, a Region of Interest （ROI） window is divided from the image. An advanced alternative of the HSV-color filter is applied on every pixel to see whether it belongs to the line. Secondly, Gaussian Smoot- hing is applied to remove small distortions in the ROI. Finally, a Probabilistic Hough Transform （PHT） is performed to extract the line from the pre-processed image and determine a directional vector for making the platform move accordingly. Test results show that efficiency and ensure an accurate line-following task. this method can effectively improve the line detection
出处
《四川兵工学报》
CAS
2014年第11期106-109,共4页
Journal of Sichuan Ordnance
