摘要
近年来移动终端的普及促进了移动平台上图像检索技术的发展。当用户看到感兴趣的商品的时候,他们希望能够使用终端拍下来,然后进行商品的检索并返回一些推荐的商家。为了解决这个问题,面向移动平台,构建了一个图像检索系统,通过手机等移动终端,拍摄或传输图片来检索互联网上相关的图片和信息。该系统构建了一个爬虫系统用来采集图片信息,在安卓平台上直接进行图像特征提取,通过移动终端拍摄的商品图像搜索互联网图像,返回相关网店链接并进行相关商品推荐。该系统对120万幅图片采用位置敏感哈希索引、存储和检索,既保证了结果在较小的误差范围内,也极大地降低了时间复杂度。最后用户可以根据推荐的链接进行选购。实验结果表明,该系统能够满足用户的需求,并且具有很强的实用性。
The popularity of mobile device in recent years promotes the development of image retrieval technology in mobile platform. When users see some interesting commodities,they hope to take a photo about them and retrieve them to get some recommended shops. In order to solve this problem, a novel image retrieval system is built facing mobile platform. By taking or transferring picture with mobile platform like mobile phones,the relevant web or information including the similar picture with the proposed system is retrieved. In the proposed system, the image featt-res are extracted on the Android platform. The web links including the similar image are returned, and recommends are given. In order to collect images, a crawling system is established. For 1.2 million images, the proposed system adopts Location Sensitive Hash to index and store. The proposed system not only promotes the retrieval performance, but also reduces the time complexity greatly. Users can buy commodities according to the recommended links. The experimental results show the system can meet users＇ needs and has very strong practicalities.
出处
《计算机技术与发展》
2016年第11期10-13,共4页
Computer Technology and Development
基金
国家自然科学基金资助项目(61202227)
安徽省自然科学基金(1408085MF122
1508085MF127)
安徽大学大学生科研训练计划项目(J10118520149)
安徽大学信息保障技术协同创新中心公开招标课题(ADXXBZ2014-5
ADXXBZ2014-6)
关键词
特征提取
图像检索
图像搜索引擎
爬虫系统
feature extraction
image retrieval
image search engine
spider system
