An Improved Ensemble-learning-based CBIR Algorithm

摘要

Traditional Content-based Image Retrieval (CBIR) algorithms are based on low-level features of images, which leads to a big margin for improvement in retrieval performance. To solve this problem, we propose a two-stage CBIR algorithm in the paper. Firstly, considering the strong ability of Convolutional Neural Network (CNN) in feature extraction, CNN-based models are established to extract high-level features for image retrieval. Secondly, Ensemble Learning (EL) framework is employed to form a new CBIR algorithm. Finally, experiments are implemented to compare the performance of the proposed algorithm with traditional algorithms. The results show that our algorithm has better image retrieval capability and stronger generalization ability.