摘要
Vapnik等学者首先提出了实现统计学习理论中结构风险最小化原则的实用算法——支持向量机,成功地解决了模式分类问题。支持向量机是目前车牌识别领域常用的算法之一,但由于实际获取的车牌图像往往存在大量的噪声干扰,大大影响了识别率。因此着眼于研究支持向量机对含噪声图片的识别效果,以字符识别为例进行分析,并与BP神经网络算法作对比,实验证明支持向量机具有较好的抑制噪声能力。
Vapnik and his collaborators proposed a useful algorithm： support vector machine, which could implement the structural risk minimization principle in statistical learning theory. Support vector machine is a useful algorithm in the practical use of license plates recognition, But the real images oflicense plates usually contain lots ofnoisy disturbance factors, which greatly influence the recognition rate, So the resistance to noisy input of support vector machine in character recognition is found out. Character recognition is taken as an example to do this research, Furthermore, the support vector machine algorithm is compared with the back propagation neural network algorithm to see which one is the best, Finally, the conclusion could be safely reached that the support vector machine has a high accuracy rate when recognizing character images with noisy input.
出处
《计算机工程与设计》
CSCD
北大核心
2007年第16期3963-3964,F0003,共3页
Computer Engineering and Design
基金
天津市科技发展计划基金项目(04310951R)
关键词
支持向量机
BP神经网络
识别率
噪声
字符识别
support vector machine
neural network
recognition rate
noisy input
character recognition
