Image retrieval based on interactive differential evolution

Different from text-based image retrieval, the content-based image retrieval (CBIR) uses low-level visual features to retrieve images. The question of how to reduce semantic gap between the low level visual features and the high level image semantics is still a difficult problem. This paper uses a comparison-based mechanism based on interactive differential evolution (IDE) to help users retrieve their preferred images in a user-oriented way. The effect of the proposed framework is evaluated, and the performance of the technique is better than that of the relevance feedback (RF) based on feature re-weighting method.