基于边缘增强型MSER特征的道路标志检测

Traffic sign detection in natural images with edge-enhancement MSER

针对自然场景下道路标志检测所面临的一些问题,提出了一种基于边缘增强型MSER特征的标志检测方法。首先采用灰度世界法对图像做光照平衡处理,并对处理后的图像进行颜色增强,区分标志和环境背景;然后基于边缘增强型MSER特征,提取标志候选区域;最后对这些候选区域使用基于霍夫变换的形状分析方法进行筛选处理。通过在GTSDB数据集上的实验验证,该方法对光照条件、局部遮挡、旋转尺度变化等情况均具有较好的鲁棒性。

Aiming at the various problems which occur in the process of traffic sign detection in natural images, a novel traffic sign detection method based on edge-enhancement MSER（Maximally Stable Extremal Regions） feature is proposed. Firstly, the gray world balance method is used for the image preprocessing to reduce the effect of illumination change, and through color enhancement, the traffic signs are distinguished from the environment background. Then, the traffic sign ROI candidate regions are extracted with edge-enhancement MSER feature. Finally, these candidate regions are further filtered by using a shape analysis method based on Hough transform. The experiment results on GTSDB（German Traffic Sign Detection Benchmark） data sets show that the proposed method is robust to lighting condition, partial occlusion, and rotation scale change.