A Fusion of Bag of Word Model and Hierarchical K-Means++ in Image Retrieval

This paper proposes an Image Retrieval method using the Bag of Word model combining with Hierarchical K-Means++ algorithm to store and arrange databases, called BoW-HKM++ approach. A Hierarchical tree is proposed to improve the processing speed in retrieval data. In this research: (1) we first build a bag of word model based on SIFT feature extraction on images; (2) then show its improvement by using inverted index technical for whole feature databases; (3) the Hierarchical K-Means++ algorithm is also proposed in the bag of word model to optimize retrieval system. Experiment results were evaluated with k-fold cross-validation method on 2 datasets of James Z. Wang Research Group. Finally, the obtained results will be compared with other method and discussed to prove the effectiveness of proposed framework in Image Retrieval problems.