摘要
图像搜索时需要尽可能地保留图像语义相似性,传统的哈希图像检索方法大多是基于人工视觉特征的,它的性能已经满足不了现在图像搜索的要求.为了解决这个问题,我们将哈希编码和卷积神经网络结合起来,旨在研究出一个更好的检索方法.本文使用卷积神经网络中的AlexNet模型和哈希编码结合,改进了传统的图像搜索算法,缩短了神经网络的训练时间,提高了哈希算法的效果.我们选用CIFAR-10数据集来进行相关实验.实验结果表明,该方法相比于传统的图像检索方法具有多方面的优越性.
Image retrieval needs to preserve image semantic similarity as much as possible. Traditional hash image retrieval methods are mostly based on artificial visual features, and its performance can not meet the requirements of current image search. To solve this problem, we combine hash coding and convolutional neural network to develop a better retrieval method. This paper uses the combination of AlexNet model of convolutional neural network and hash coding,which improves the traditional image retrieval algorithm and the effect of hash algorithm and shortens the training time of neural network. We selected the CIFAR-10 data set to carry out related experiments. The experimental results show that the method has many advantages compared with the traditional image retrieval method.
作者
衣姝颖
白璐
李天平
Yi Shuying;Bai Lu;Li Tianping(School of Physics and Electronics,Shandong Normal University,250358,Jinan,China)
出处
《山东师范大学学报（自然科学版）》
CAS
2019年第1期88-95,共8页
Journal of Shandong Normal University(Natural Science)
关键词
哈希编码
卷积神经网络
图像检索
Hash coding
convolutional neural network
image retrieval
