【摘要】 提出一种快速、实用的基于Gabor滤波器的车型识别方法。该方法包括车辆分割、特征提取和模板匹配3个阶段。首先在车辆分割阶段,采用基于对数密度的背景消减法,降低光照变化带来的影响;其次提出一种新的非均匀采样策略,提取Gabor特征;最后应用模板匹配的方法对车型进行识别。和传统方法相比,该方法能够有效降低Gabor滤波器的计算量和存储空间,同时又使识别率和鲁棒性得到明显增强。实验数据亦表明了该方法的可行性和有效性。更多还原

【Abstract】 In this paper,a fast and practical vehicle classification method based on Gabor features is proposed,which contains three consecutive stages: Vehicle segmentation,Gabor features extraction and template matching.In the first stage,background subtraction is applied based on logarithmic intensities instead of standard intensities to weaken the affection of illumination change.In the second stage,it puts forward a novel non-even sampling of Gabor features for classification on the basis of the edge features in vehicles.Finally,the vehicle classification is done by means of template matching.Comparing with the conventional method,it heavily reduces the computation and memory requirements,and illustrates good performance both in discrimination ability and robustness.The experimental data show that the method proposed here is available and efficient.更多还原