基于MSER和PHOG特征的交通标志检测方法

Traffic Sign Detection Based on MSER and PHOG Feature

本文提出了一种基于最大稳定极值区域(Maximally Stable Extremal Regions,MSER)和塔式梯度方向直方图(Pyramid Histogram of Oriented Gradients,PHOG)特征的交通标志检测方法。该方法首先对图像进行分通道颜色增强,再利用最大稳定极值区域算法进行交通标志潜在区域的定位和提取,然后通过提取目标图像的PHOG特征,结合支持向量机(Support Vector Machine,SVM)训练形状分类器进行交通标志的粗分类。实验结果表明,该方法可以有效地抑制光照、遮挡以及场景复杂等因素带来的影响,并获得了较高的检测率及较低的误检率,同时也为后续的标志识别工作打下基础。

We propose a detection method for traffic signs based on the maximally stable extremal regions （MSER） and pyramid histogram of oriented gradients （PHOG） features. In this method, the image color is firstly enhanced in sub-channel, using MSER algorithms locates and extracts the potential areas of traMc signs. Then the PHOG features of the target image is extracted, and combined with support vector machine training shape classifier, coarse classification of traffic signs is implemented. Experimental results show that this method is robust on the factors of illumination, occlusion and complex background, and gets a higher detection rate and low false positive rate, but also can lay the foundation for the subsequent work of sign recognition.