Deep Multi-View Enhancement Hashing for Image Retrieval

Hashing is an efficient method for nearest neighbor search in large-scale data space by embedding high-dimensional feature descriptors into a similarity preserving Hamming with low dimension. However, high-speed retrieval through binary code has certain degree of reduction accuracy compared to traditional methods. We have noticed that multi-view methods can well preserve the diverse characteristics data. Therefore, we try introduce deep neural network hash learning field, and design innovative model, which achieved significant improvement performance. In this paper, propose supervised model enhance information networks. This completely new combines The proposed utilizes effective view stability evaluation actively explore relationship among views, will affect optimization direction entire network. also designed variety multi-data fusion advantages both convolution multi-view. order avoid excessive computing resources on enhancement procedure during retrieval, set up separate structure called memory participates training together. systematically evaluated CIFAR-10, NUS-WIDE MS-COCO datasets, results show our significantly outperforms state-of-the-art single-view hashing