针对自然场景图像中交通标志存在多视角难以识别的问题,提出了一种基于ASIFT(affine-scale-in-variant feature transform)算子匹配的交通标志识别算法。利用ASIFT对视角变化的不变性进行特征点匹配,通过统计并比较有效匹配点来决定标志种类。实验表明,该算法对视角较小的交通标志也有较好的效果。
Considering the multi-view traffic signs in the nature scenario.This paper proposesa ASIFT-MATCH(affine-scale-invariant feature transform,ASIFT)based multi-view trafficsign recognition method.Making use of the ASIFT invariant features,we classify the signsby calculating the number of the valid matching points.Experiments show this method perform well even on the low-angle signs.