An image retrieval method based on multi-instance learning and perturbative glowworm swarm optimization

作者：Chen, Tao^*; Deng, Huifang; Dong, Chunru

来源：Journal of Information and Computational Science, 2013, 10(10): 2951-2959.

DOI：10.12733/jics20101872

• Function Optimization
• Global optimal solutions
• Glowworm swarm optimizations
• Multi-instance learning
• Object-based image retrieval
• Similarity measurements
• Swarm intelligence algorithms
• User's relevance feedback

For being able to identify ambiguous object effectively, Multi-instance Learning (MIL) has been applied to deal with various real problems such as the object-based image retrieval, which is essentially a problem of function optimization. To improve the performance of solving such an optimization function, in this paper, we proposed a novel image retrieval method based on MIL and Perturbative Glowworm Swarm Optimization (PGSO). The reason for adopting the Glowworm Swarm Optimization (GSO) is that it has the characteristics of high precision and fast convergence in searching a global optimal solution. The process of MIL-PGSO mainly includes the following three steps: First, it segments each image into a number of regions, treats images and regions as bags and instances respectively, and then an objective function of multi-instance learning is constructed;Second, it uses PGSO algorithm to search the user desired target concept, which is also the global optimum of the objective function;Finally, according to similarity measurement and user's relevance feedback, it returns a set of satisfying images to user. Experimental results on COREL image data sets show that the proposed approach has high retrieval accuracy and low time consuming.

• 出版日期2013-7-1
• 单位华南理工大学