【摘要】 为了准确地检测交通标志,需要将汽车智能辅助驾驶系统拍摄到的倾斜图像矫正为无倾斜的标准图像。提出了一种基于仿射变换的三角形交通标志矫正方法,根据三角形交通标志共有的形状特性,提取出交通标志的边缘,求出三角形三边所在直线方程的交点及三条边的中点共六个特征点的坐标,与一个预先设定的标准正三角形的顶点及各边中点相对应,通过仿射变换将倾斜的交通标志矫正,得到一个与标准三角形交通标志仅差一个缩放因子的标准图像,为后续的交通图像识别打下基础。更多还原

【Abstract】 In order to detect the traffic signs accurately, a tilt image of the traffic sign captured by an intelligent driving assistance system should be converted to a standard image without tilt and rotation. A correcting method for triangle traffic signs based on affine transformation was proposed. The edge of triangular traffic sign was extracted, and the three straight lines outermost of edges and their mathematical equations were figured out. Therefore, three intersections, as well as per midpoint of side which composed by three lines should be solved. Another equilateral triangle was set to correspond to the first one, and coordinates of six points, including three intersections and three midpoints were known. These twelve points, called as feature points, compose six feature point-pairs. As a result, the tilt image can be corrected by affine transformation with these six feature point-pairs, and the image obtained has just a scale factor compared with the real triangle traffic sign. It lays a good foundation to the following traffic image recognition.更多还原