基于HOG-Gabor特征融合与Softmax分类器的交通标志识别方法

Traffic sign recognition method based on HOG-Gabor feature fusion and Softmax classifier

为了提高交通标志识别的正确率和实时性,提出了一种基于HOG-Gabor特征融合与Softmax分类器的交通标志识别方法。采用Gamma矫正方法提取HOG特征,采用对比度受限的自适应直方图均衡化方法提取Gabor特征,基于线性特征融合原理,将提取的HOG和Gabor特征向量直接串联,得到刻画交通标志的融合特征向量,采用Softmax分类器对融合特征向量进行分类,采用德国交通标志识别基准(GTSRB)数据库测试了所提方法的有效性,比较了基于单特征与融合特征的交通标志识别效果。试验结果表明:在图像增强过程中,针对HOG特征,采用Gamma矫正方法的分类正确率最大,为97.11%,针对Gabor特征,采用限制对比度的直方图均衡化方法的分类正确率最大,为97.54%;采用Softmax分类器的最小分类正确率为97.11%,耗时小于2s;针对HOG-Gabor融合特征,采Softmax分类器的识别率高达97.68%,因此,基于HOG-Gabor特征融合与Softmax分类器的交通标志识别方法的识别率高,实时性强。

In order to improve the accuracy and real-time performance of traffic sign recognition,a traffic sign recognition method was proposed based on HOG-Gabor feature fusion and Softmax classifier.HOG（histogram of oriented gradient）feature was extracted by using the Gamma correction method,and Gabor feature was extracted by using the contrast limited adaptive histogram equalization method.According to the linear feature fusion principle,HOG and Gabor feature vectors were connected to constitute the fusional feature vector for depicting the traffic signs.Theeffectiveness of the proposed method was verified based on the GTSRB（German Traffic Sign Recognition Benchmark）data set.The recognition effects of traffic sign based on single feature and fusional feature were compared.Experimental result shows that in image enhancement,the classification accuracy based on HOG feature is 97.11% and is largest by the Gamma correction method,and the classification accuracy based on Gabor feature is 97.54% and is largest by the contrast limited adaptive histogram equalization method.The minimum classification accuracy is 97.11% by using Softmax classifier,and classification time is only 2s.The correct recognition rate of traffic sign reaches 97.68% by using the proposed method based on HOG-Gabor fusional features,so the traffic sign recognition method based on HOG-Gabor fusional features and Softmax classifier has high recognition rate and real-time performance.