【摘要】 为了解决在复杂场景中进行车牌定位的问题,提出了一种基于MSER与DRLBP特征的车牌定位方法。首先对输入图像进行预处理,然后在多个通道上进行MSER候选区域提取;接着利用所设计的基于车牌字符合并的车牌定位方法进行车牌字符合并;最后利用DRLBP纹理特征对合并后的区域进行验证从而得到最终的车牌区域。实验结果表明该方法具有较好的定位能力。由于方法是通过寻找车牌字符进而定位车牌位置,因此其受车牌颜色、车牌格式的影响较小,在复杂环境中对国内外不同车牌均有较好定位效果。更多还原

【Abstract】 To solve the problem of license plate location in complex scenes, a license plate localization method was proposed based on MSER and DRLBP features. First, the input image was preprocessed, and then the MSER algorithm was used to extract possible license plate character fields in multiple channels. Secondly, a new method for locating the license plate position was implemented. The main idea of this method is to merge the license plate character fields into the full license plate, and to locate the license plate. Finally, the DRLBP texture feature was utilized for examining the results of the last step. And then, the final results were generated. The experimental results show that the method was very effective. Because this method is to find the characters of license plate and locate the location of license plate placed, it is less affected by the color and format of license plate.更多还原