Reverse image search for scientific data within and beyond the visible spectrum
作者：
Highlights：
• Content-based image retrieval tool ranks scientific pictures by similarity quickly.
• Recommendation system enables people without specialized knowledge to search images.
• ConvNets and locality sensitive hashing forest speed up image retrieval.
• Tested on four databases ranging from thousands to millions samples.
• Comparisons among ConvNet architectures and science domains.
摘要
•Content-based image retrieval tool ranks scientific pictures by similarity quickly.•Recommendation system enables people without specialized knowledge to search images.•ConvNets and locality sensitive hashing forest speed up image retrieval.•Tested on four databases ranging from thousands to millions samples.•Comparisons among ConvNet architectures and science domains.
论文关键词：Reverse image search,Content-based image retrieval,Scientific image recommendation,Convolutional neural network
论文评审过程：Received 21 December 2017, Revised 2 April 2018, Accepted 11 May 2018, Available online 25 May 2018, Version of Record 25 May 2018.
论文官网地址：https://doi.org/10.1016/j.eswa.2018.05.015
