图像处理技术在车牌号识别领域中的算法改进

针对图像处理技术在车牌号识别领域中的运用,结合目前常用识别方法(模板匹配法、神经网络法和支持向量机识别法)和主要特征提取方式(统计特征提取和结构特征提取)各自的优点,设计出一种多层分支结构的车牌号识别系统,该系统根据不同待识别字符的特征进行分类,然后匹配上对该类特征识别较有优势的特征提取方法和字符识别算法,并提出一种对折识别算法运用其中,通过对字符的分层识别和分支识别,从而达到精确、高效识别的目的,最后通过试验测试和统计分析,证明了该方法在车牌号识别中的优越性.

The improvement of image processing algorithm in license plate recognition

Abstract: This paper introduce a design of license plate number recognition system with multi layer branch structure,which combines the advantage of template matching method, neural network method , support vector machine method, statistical feature extraction and structure feature extraction method. Firstly, this system classify these characteristics into different classes, then match these classes with feature extraction and character recognition algorithm,and a fold recognition algorithm is also employed in it. With the help of the characters and branch recognition, the system performance achieve accurate and efficient identification. Finally, the experiment results and statistical analysis prove that the design perform better than other systems in the license plate number recognition.