Cross-Modal Hashing For Efficient Multimedia Retrieval

The explosive growth in big data has attracted much attention in designing efficient indexing and search methods recently.In many critical applications such as large-scale search and pattern matching,finding the nearest neighbors to a query is a fundamental research problem.However,the straightforward solution using exhaustive comparison is infeasible due to the prohibitive computational complexity and memory requirement.In response,Approximate Nearest Neighbor(ANN)search based on hashing techniques has become popular due to its promising performance in both efficiency and accuracy.Since heterogeneity has been an increasingly important characteristic,numerous cross-modal retrieval methods have been proposed.Cross-modal hashing also has attracted considerable attention.Recently,more and more sophisticated researches related to this topic are proposed.However,they also have several problems.First,they seem to be inefficient computationally for several reasons.On one hand,learning coupled hash projections makes the iterative optimization problem challenging.On the other hand,individual collective binary codes for each modality are also learned with a high computation complexity.Second,since learning binary codes is a challenging integer optimization problem,most existing cross-modal hashing methods aim to map heterogeneous data into one common low-dimensional hamming space and then threshold to obtain binary codes.However,this independent relaxation step brings bad quantization quality,resulting in poor retrieval performances.Third,many methods are effective for one of signal-label dataset,multi-label dataset,large-scale datasets or high-dimension dataset.But there are few methods can apply to all the datasets.Base on the proposed challenges,in this paper,we focus on how to design hashing functions and optimization methods.Specific work is as follows.First,we describe a simple yet effective cross-modal hashing approach that can be implemented in just three lines of code.This approach first obtains the binary codes for one modality via unimodal hashing methods then applies simple linear regression to project the other modalities into the obtained binary subspace.Obviously,it is non-iterative and parameter-free,which makes it more attractive for many real-world applications.Second,we propose a novel supervised cross-modal hashing method called Discrete Cross-Modal Hashing(DCMH)to learn the discrete binary codes without relaxing them.DCMH is formulated through reconstructing the semantic similarity matrix and learning binary codes as ideal features for classification.Furthermore,DCMH alternately updates binary codes of each modality,and iteratively learns the discrete hashing codes bit by bit efficiently.Third,how to design the hashing function is very important for cross-modal hashing methods.However,many published methods have shown that optimization method is also play a key role for retrieval performance.Therefore,we also optimize our objective function in a relax-and-threshold manner to evaluate the effectiveness of proposed discrete optimization.