分布式平台下自适应的快速图像检索

With the popularity of mobile phones and other photographic devices, image data is growing rapidly. How to carry out effective image retrieval is still an urgent problem to be solved. Aiming at the problems of slow speed and low difficult expansion of traditional single-machine image retrieval, this paper proposes a parallel retrieval method based on Hadoop for image retrieval task. In addition, in order to avoid the difficulty of storing image features, this paper uses Hbase to store image features. Because the deep image features can better present the semantic information of the whole image, this paper uses deep image features based on CNN for image retrieval task.The experiments on Flickr1M and UKbench datasets show that the efficiency and accuracy ofimage retrieval task are improved based on our method .