Share Email Print

Optical Engineering

Multi-clues image retrieval based on improved color invariants
Author(s): Liu Liu; Jian-Xun Li
Format Member Price Non-Member Price
PDF $20.00 $25.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

At present, image retrieval has a great progress in indexing efficiency and memory usage, which mainly benefits from the utilization of the text retrieval technology, such as the bag-of-features (BOF) model and the inverted-file structure. Meanwhile, because the robust local feature invariants are selected to establish BOF, the retrieval precision of BOF is enhanced, especially when it is applied to a large-scale database. However, these local feature invariants mainly consider the geometric variance of the objects in the images, and thus the color information of the objects fails to be made use of. Because of the development of the information technology and Internet, the majority of our retrieval objects is color images. Therefore, retrieval performance can be further improved through proper utilization of the color information. We propose an improved method through analyzing the flaw of shadow-shading quasi-invariant. The response and performance of shadow-shading quasi-invariant for the object edge with the variance of lighting are enhanced. The color descriptors of the invariant regions are extracted and integrated into BOF based on the local feature. The robustness of the algorithm and the improvement of the performance are verified in the final experiments.

Paper Details

Date Published: 4 May 2012
PDF: 9 pages
Opt. Eng. 51(5) 057003 doi: 10.1117/1.OE.51.5.057003
Published in: Optical Engineering Volume 51, Issue 5
Show Author Affiliations
Liu Liu, Shanghai Jiao Tong Univ. (China)
Jian-Xun Li, Shanghai Jiao Tong Univ. (China)

© SPIE. Terms of Use
Back to Top