摘要
由于当前方法存在识别率较低、识别效果不理想等问题,提出基于离散余弦变换的电力营销系统客户权限自动识别方法,该方法首先利用PCA方法对所选向量进行降维处理,并利用LDA特征提取方法通过线性变换提取人脸特征,最大化分离所获得的特征;其次通过DDCT特征提取方法对图像进行变换处理,利用Zig-zag获得特征提取过程中所需的特征向量以及低频分量,再融合LDA与DDCT两种方法提取人脸特征,最后构建一个支持向量机的分类器,将所提取的特征输入其中,实现客户权限自动识别的操作。实验结果表明,所提方法的识别率以及识别效果较好,能够保证识别率。
Due to the low recognition rate and unsatisfactory recognition effect of the current method,an automatic recognition method of power marketing system customer authority based on discrete cosine transform is proposed.The method first uses the PCA method to reduce the dimension of the selected vector,and uses the LDA feature extraction method to extract facial features through linear transformation to maximize Separate the obtained features;secondly,it uses the DDCT feature extraction method to transform the image,uses Zig-zag to obtain the feature vectors and low-frequency components required in the feature extraction process,and then merges the LDA and DDCT two methods to extract the facial features,and finally constructs a support vector machine classifier,inputs the extracted features into it,and realizes the operation of automatic recognition of customer authority.The experimental results show that the recognition rate and recognition effect of the proposed method are better,which can guarantee the recognition rate.
作者
胡美慧
向志威
HU Mei-hui;XIANG Zhi-wei(State Grid Xin Jiang Information&Telecommunication Company,Urumchi 830000 China)
出处
《自动化技术与应用》
2022年第5期125-129,共5页
Techniques of Automation and Applications
关键词
自动识别方法
离散余弦变换
特征识别
DDCT
支持向量机
automatic identification method
discrete cosine transform
feature identification
DDCT
support vector machine
