摘要
通过改进处理K-SVD算法得到DDELM-AE算法,之后再把该算法应用于物体特征识别中。研究结果得到:K-SVD算法的收敛速率较快,达到收敛的时间也显著比SAE算法更短,本文通过改进K-SVD算法之后使DDELM-AE算法的计算准确率以及计算效率都获得了显著的改善。采用K-SVD算法可以达到76.3%的识别准确率,使用深度特征信息之后,可以使识别准确率升高至81.4%,DDELM-AE可以显著提高K-SVD算法的性能,并且加入多特征之后可以使算法识别准确率得到显著提高。
To The improved K-SVD algorithm is used to obtain the DDELM-A algorithm, which is then applied to object feature recognition.The research results show that the convergence rate of the K-SVD algorithm is faster and the convergence time is significantly shorter than that of the SAE algorithm. In this paper, after the improvement of the K-SVD algorithm, the computational accuracy and computational efficiency of the DDELM-A algorithm are significantly improved.The adoption of the K-SVD algorithm can achieve the recognition accuracy rate of 76.3%. After the use of depth feature information, the recognition accuracy can be increased to 81.4%. DDELM-A can significantly improve the performance of the K-SVD algorithm.
作者
殷晓辉
Yin Xiao-hui(Department of Information and Control Engineering,Yantai Automobile Engineering College,Yantai,Shandong 264000,China)
出处
《生命科学仪器》
2018年第6期46-49,36,共5页
Life Science Instruments