摘要
文章给出了自行设计的在低功耗嵌入式平台上实现车牌定位检测的算法。首先通过Sobel水平算子对车牌图像进行垂直边缘检测,运用形态学方法对边缘图像进行闭运算得到连通图块。然后根据连通区域轮廓确定最小外接矩形,解决车牌位置断节问题,得到车牌候选区域。最后根据车牌特征对提取出的车牌候选区域进行筛选实现准确的车牌定位。实验结果证明了该算法的有效性,而且在复杂背景下也具有较好的鲁棒性和自适应性。
A Chinese number plate locating algorithm is proposed for low power consumption embedded system. Sobel operator is applied in hor- izontal direction to detect vertical edge component. Connected edge outputs are obtained by image morphology operations. Minimum rectangular area is determined with the connected edge outputs. Then the possible range of number plated is obtained, its location is fine-tuned with known properties of number plate. Experimental results show that the proposed algorithm is effective and it also has good robustness and adaptability in various application environments.
出处
《微型机与应用》
2017年第3期45-48,58,共5页
Microcomputer & Its Applications
关键词
车牌定位
SOBEL算子
形态学
闭运算
number plate locating
Sobel operator
morphology
close operation
