摘要
提出了一种新的基于商品图像的检索系统,充分利用当前学术界的一些高效算法,包括基于Hadoop平台的大数据处理技术,基于E2LSH的高维数据近邻查找技术,基于图像全局特征提取的GIST技术以及基于深度学习的卷积神经网络技术CNN。紧密结合这些新技术,在基于商品图像的检索方面取得了较好的检索效果。
The paper proposed a new commodity image retrieval system, making full use of the current academic efficient algorithm, including large data processing technology based on Hadoop platform, high dimensional data nearest neighbor search technology based on E2LSH, GIST feature extraction based on global feature extraction of image, as well as the convolution neural network technology（CNN） based on deep learning. Base on these new technologies, the new system can obtain a better retrieval result.
出处
《移动通信》
2016年第8期63-69,74,共8页
Mobile Communications
基金
宁波市自然科学基金资助项目(2014A610023)
宁波市自然科学基金资助项目(2015A610119)
