摘要
针对传统的车牌定位算法是采用灰度图像的纹理特征的,由于灰度图像的纹理特征诸多特性和局限性,给车牌定位算法带来很大的困难;鉴于此,提出了基于小波纹理分析的彩色车牌定位算法;首先对图像进行二维小波分解,计算滑动窗内图像的小波纹理特征;然后同时将低频子图转化RGB色彩分量为HSV色品值,提取颜色特征并得到综合特征向量;其次利用小波变化定位出车牌区域;最后对定位出的车牌进行后期校正处理并输出定位结果;该算法采用综合特征定位,克服了采用单一特征定位的缺点和局限性,从实验结果可以看出该算法能够较准确的从背景图像中提取出车牌位置信息。
The traditional license plate localization algorithm uses the gray image texture feature, the texture feature of gray image lea tures and limitations bring great difficulty in getting a license plate localization algorithm. In view of this, put forward based on the Moire principle analysis of the color of license plate localization algorithm. Firstly, two--dimensional wavelet decomposition for image, sliding win- dow＂ is small ripple bedding features of images; And at the same time convert low--frequency subgraph RGB color components as HSV color value, extract color feature, and get a comprehensive feature vector; Second, wavelet variations positioning license plate area; Finally to locate the license plate positioning results of late correction processing and output. The algorithm adopts the comprehensive characteristics of localization, overcomes the drawback of using single feature localization and limitations, and can be seen from the results of this algorithm can accurately extract the license plate location information from the background image.
出处
《计算机测量与控制》
北大核心
2014年第1期159-161,165,共4页
Computer Measurement &Control
基金
新疆大学校院联合资助项目(XY110133)
国家自然科学基金(60865001)
关键词
车牌定位
小波纹理
颜色特征
license plate location
small ripple bedding
color features
