基于归一化RGB与椭圆相似度的圆形交通标志检测

Circular traffic sign detection based on NRGB and elliptical similarity

提出一种基于归一化RGB(NRGB)和椭圆相似度的圆形交通标志检测方法。首先将图像的RGB模型转换为NRGB,利用阈值分割得到红、蓝、黄颜色分量,然后根据面积和长宽比去除干扰区域,最后利用椭圆相似度检测感兴趣区域是否为圆型交通标志。实验结果表明,该方法能在复杂背景中准确定位圆形交通标志,获得了较高的检测率和较低的误检率。

We propose a method of traffic sign detection, which is based on normalized RGB and elliptical similarity. Firstly, RGB model is converted to normalized RGB. Then, remove the interference according to area and aspect ratio. Finally, use the elliptical similarity to detect circular traffic signs. The results indicate that the method can accurately locate the circular traffic signs, and achieve high detection rate and low false rate.