基于混合特征的多车牌定位算法

Hybrid-feature based multi-vehicle plate location algorithm

本文提出了一种基于混合特征的多车牌定位的新方法,先根据车牌图像的边缘特征,利用数学形态学的方法生成连通区域,搜索全图得到连通区域的最小外接矩形。然后利用我国车牌本身的几何特征和颜色特征进行判别,提取出车牌的候选区域并判别出车牌的颜色。交通图像处理实验证明该方法具有限制条件少,速度快和准确率高的优点,并可进行多车牌定位。

A novel algorithm for multi-vehicle plate location based on the hybrid features is put forward in the paper. Based on the edge features of the vehicle plate, the connective regions is generated with the method of morphology and the relevant minimum enclosing rectangular section can be obtained by the whole-map searching. Then the candidate regions can be picked up with the features of color and configuration Of the license plates of our country. The experimental traffic image processing verified that the proposed algorithm has high accuracy and improved speed with little limitation for muhi-vehicle plate location compared with the traditional method.