基于视觉注意模型和SIFT的交通标志识别方法

Traffic signs recognition based on visual attention model and SIFT

为了提高交通标志的识别速度和识别率,提出了一种基于视觉注意模型和SIFT特征的交通标志识别方法.首先基于视觉注意模型提取颜色特征,找出交通标志可能的候选区域,然后对候选区域进行SIFT特征提取,与标准交通标志图像库进行相似度计算,可实现快速准确的检测与识别.与传统方法相比,具有无需精确分割、计算量小、体现仿生学特性等优点.在采自国内外的两组交通标志图像库上进行交通标志识别测试,都得到了良好的效果.

In order to improve the speed and accuracy of traffic signs recognition （TSR） , a novel method based on a visual attention model and scale invariant feature transform （SIFT） feature is proposed. First, color features are extracted based on the visual attention model, and the candidate regions of traffic signs could be achieved. Then, SIFT feature of the candidate region is extracted, and the similarity calculation is done be- tween the candidate region and the standard traffic signs. Thus, the traffic sign would be recognized. The pro- posed method can recognize the traffic signs more rapidly and accurately than traditional methods, and it is characterized by no need for segmentation, less calculation, and bionics. The experiments results on two traffic signs databases collected from home and abroad show the excellent effect of the method on TSR.