Non-Parametric Quantum Theory Based Image Retrieval

Recently learning to rank has become one of the most common methods to build a ranking model for social image retrieval. However, the results of existing approaches are not so satisfactory for the large gap between low-level visual features and high-level semantic concepts, and these approaches require a significant amount of parameters tuning in the design process to be effective and efficient. In this paper, we propose a novel framework for social image retrieval based on a non-parametric quantum theory, which ranks images by considering their inter-relationship through the quantum estimation without explicit parameter tuning. The basic idea of the proposed framework is inspired by the photon polarization experiment that supports the theory of quantum measurement. Experimental results conducted on the Corel dataset demonstrate the effectiveness and efficiency of the proposed framework.