Efficient target search with relevance feedback for large CBIR systems

Abstract

Recent content-based image retrieval (CBIR) techniques were designed around query refinement based on relevance feedback. They suffer from slow convergence, high disk I/O, and do not even guarantee to find intended targets. In this paper, we identify the cause of these problems and propose several efficient target search methods to address these drawbacks. Our complexity analysis shows that our approach is able to reach any given target image with fewer iterations in the worst and average cases. We evaluated our techniques on large datasets in simulated and realistic environments. The results show that our approach significantly reduces the number of iterations and improves overall retrieval performance. The experiments also confirm that our approach can always retrieve intended targets even with poor selection of initial query points and can be used to improve the effectiveness of existing CBIR systems with relevance feedback.

Intelligent Image Retrieval Techniques: A Survey

“…Efficient schemes for image retrieval with smart logic and understanding are presented in [67][68]. The scheme proposes the limited iteration model for accommodating the user's feedback as this can be trivial on large image databases; hence, as alternative other schemes are used to extract the user's intension instead of iteratively recording the user's response.…”

Query Decomposition: A Multiple Neighborhood Approach to Relevance Feedback Processing in Content-based Image Retrieval

“…The design of the above and some other existing CBIR techniques, e.g., [2,8,10,15,20,21], are based on the k-NN model: objects of similar semantics look similar in many aspects. This strategy fails when the image characterization and similarity measure do not follow perceptual characteristics.…”

Adequando consultas por similaridade para reduzir a descontinuidade semântica na recuperação de imagens por conteúdo

“…Em seguida, o novo centróideé calculado, levando em consideração o grau de relevância de cada imagem, e então eleé utilizado para selecionar as imagens na consulta refinada. Essa estratégia foi utilizada em [Rui et al, 1998, Ishikawa et al, 1998, Liu et al, 2006a, Doulamis e Doulamis, 2006, Liu et al, 2006b, Shen et al, 2009.…”

“…Em [Liu et al, 2006a, Liu et al, 2006b] são apresentadas técnicas de movimentação do centro de consulta que permitem a convergência em um menor número de iterações do laço de realimentação de relevância. Elas são baseadas em amostragem aleatória e na análise da vizinhança local por meio da construção de diagramas de Voronoi para a realização da nova consulta aos vizinhos mais próximos.…”