摘要
在基于小波边缘检测和SMO算法的基础上进行车型图像识别,首先将待识别目标进行二维小波分解,获取不同尺度下的小波系数,然后对其进行主元分析,得到的主元作为支持向量机的特征量输入。实验结果表明,该方法具有良好的分类性能。
Vehicle image recognition is conducted based on edge detection by wavelets and SMO algorithm. First 2-D wavelet is used to decompose the image and get the wavelet coefficients, then calculate principal components which will be the inputs of SVM. Experiment results indicate that this method has good classification performance.
出处
《交通标准化》
2007年第11期117-120,共4页
Communications Standardization
关键词
车型识别
支持向量机
序贯最小优化
小波变换
主元分析
vehicle recognition
support vector machines
sequential minimal optimization
wavelet transform
principal components analysis
