基于小波核LS-SVM的车牌字符识别算法研究

Research on Plate Character Recognition Based on Wavelet Kernel LS-SVM

字符识别是整个车牌识别系统至关重要的一步,决定着系统最终的识别率。文中不同于传统的SVM识别方法,而是采用了LS-SVM为基础的新颖方法,从而简化了SVM优化问题的求解。鉴于车牌字符的独特性,将小波函数作为LS-SVM的核函数。结合字符和字符识别的特征,分析小波核函数的可行性,最后通过实验结果横向、纵向对比,得出小波核函数的优势。实验结果表明,相比于传统的神经网络和模板匹配等字符识别算法,提高了车牌系统的识别率;与传统SVM识别算法相比,亦减少了车牌的识别时间。

Plate character recognition is a most important step of the whole plate recognition system,which determines the final system recognition rate. The method LS-SVM used in this paper is different from the traditional SVM,which simplifies the SVM optimization.In view of the specialty of plate character,take the wavelet function as the kernel function for LS-SVM. Combined with the feature of character and character recognition,analyze the feasibility of this wavelet kernel function,finally through the experimental results of vertical and horizontal comparison,the advantage of wavelet kernel function is obtained. Compared with other algorithms such as Neural Network and Template M atching,this method has improved the recognition rate while the recognition time is reduced.