基于分块聚类的车牌图像二值化方法

利用车牌图像中车牌部分所占比例小、车牌字符与底色灰度对比度大的特点,为运算简单,易识别定位,提出了一种基于分块聚类的车牌图像的二值化方法。方法先将整幅灰度车牌图像等分成多个小块,对每个小块中的像素点根据其灰度值进行聚类,再将聚类结果二值化而得到各小块的二值化结果,合成各小块的二值化结果得到整幅车牌图像的二值图像。实验结果表明,与OTSU二值化方法相比,不仅经该方法处理得到的二值图像能很好地体现车牌特征、有效地降低车牌图像背景的复杂性,而且方法更适用于处理非正常光照条件下的车牌图像。在此基础上能有效地提高车牌定位系统的实时性和适应性。

Binary Method of License Plate Image Based on Blocking-Clustering

Making use of the features of license plate taking a small part of the hole license plate image and the stronger contrast of the characters with the background,this paper proposes a binary method of vehicle license plate image based on blocking-clustering.After dividing the whole image into n blocks,clustering is carried out on pixels according to their gray-scale level in each block,then binarization is performed according to the result of clustering,and all blocks' binarization results are combined to for...