Feature integration analysis of bag-of-features model for image retrieval

One of the biggest challenges in content based image retrieval is to solve the problem of "semantic gaps" between low-level features and high-level semantic concepts. In this paper, we aim to investigate various combinations of mid-level features to build an effective image retrieval system based on the bag-of-features (BoF) model. Specifically, we study two ways of integrating the SIFT and LBP descriptors, HOG and LBP descriptors, respectively. Based on the qualitative and quantitative evaluations on two benchmark datasets, we show that the integrations of these features yield complementary and substantial improvement on image retrieval even with noisy background and ambiguous objects. Two integration models are proposed: the patch-based integration and image-based integration. By using a weighted K-means clustering algorithm, the image-based SIFT-LBP integration achieves the best performance on the given benchmark problems comparing to the existing algorithms.