摘要
在识别过程中,首先将人脸图片分为神经元层次网络,再利用数组矩阵变换提取图片特征组合为多维向量作为整幅人脸图片特征。在此基础上,为每个类构造一个支持向量机进行识别。实验仿真表明,算法实现较简单,结果比较形象。
During face recognizing, the face pictures will be divided into neuron-level network. Then it extracts picture features using array matrix transformation. And it composes those eigenvalues into multidimensional vector to describe the face picture characteristics. Based of this, it constructs a support vector machine each category for identification. The simulation result shows that this algorithm is simple and available.
出处
《电子测量技术》
2008年第7期32-34,共3页
Electronic Measurement Technology
关键词
单层神经元网络
多层神经元网络
奇异值分解
细胞数组
single layer nerve basic network
multilayer nerve basic network
decomposing single values
cell number group
