¹China Information Technology Security Evaluation Center, Beijing, 100085, P.R. of China
²College of Information science and Technology, Beijing University of Chemical Technology, Beijing, 100029, P.R. of China

---

Abstract— In this paper, a second-order Hash retrieval approach is proposed based on SIFT feature of pictures and applied to search similar images. Firstly, extract features of an image by the method of SIFT. Then, cluster the key words through K-Means algorithm and create a word frequency table of the features by utilizing bag ofword algorithm. Finally, match familiar images by the method of second-order Hash retrieval algorithm based on the word frequency table. The second-order Hash retrieval algorithm includes two steps. The first-order Hash retrieval aims to searchsimilarities of feature distribution structure. And the second-order Hash retrieval implements accurate search, which depends on the ratio ofthe quantity of the two images’ features belong to the same feature category tothe total feature points of the image itself. The experiment results indicatethat this approach performs well on accuracy and efficiency.

Index Terms—SIFT features,image Hash retrieval,word frequency table,second-order retrievalc

[PDF]

Cite: Xianghui Zhao, Zhirong Li, and Junkai Yi, "SIFT Feature-Based Second-Order Image Hash Retrieval Approach," Journal of Software vol. 13, no. 2, pp.103-116, 2018.