摘要
车辆图像的预处理效果的好坏直接影响着车牌定位的准确率和运算速度;为了去除复杂背景区域的影响,消除图像的噪声,提出了一种针对视频序列改进的预处理流程;基于混合高斯背景建模结合小波包多尺度去噪的思想,去除大量的背景信息,缩小了车牌定位的范围再继续后面的预处理工作;上述方法得到的图片,噪声干扰降低且含有较少的不规则连通区域,分别以文献[6]和文献[7]提供的车牌定位方法为例进行测试,效果良好。
The preprocessing of a vehicle Image is an important prerequisite for license plate locating, it＇ s effect directly impact on the accuracy of license plate location and speed of operation. In order to remove the complex background of the region, eliminating image noise, this paper presents an improved preprocessing process for video sequences. Based on Gaussian mixture background modeling and the method of wavelet packet multiseale denoising. Removal of a large number of background information to narrow the scope of the license plate location and then continue to the next work. The experimental tests show that the pictures obtained by this method have less background and noise, it so that it can make a good foundation for the subsequent license plate location work.
出处
《计算机测量与控制》
CSCD
北大核心
2011年第10期2515-2517,2521,共4页
Computer Measurement &Control
基金
国家自然科学基金(60971048)
