A hybrid feature model for seam carving detection
Document Type
Conference Proceeding
Publication Date
1-1-2017
Abstract
Seam carving, as a content-aware image resizing algorithm, is widely used nowadays. In this paper, an advanced hybrid feature model is presented to reveal the trace of seam carving, especially seam carving at a low carving rate, applied to uncompressed digital images. Two groups of features are designed to capture energy variation and pixel variation caused by seam carving, respectively. As indicated by the experimental works, the state-of-the-art performance on detecting 5% and 10% carving rate cases has been improved from 81.13% and 90.26% to 85.75% and 94.87%, respectively.
Identifier
85028464730 (Scopus)
ISBN
[9783319641843]
Publication Title
Lecture Notes in Computer Science Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics
External Full Text Location
https://doi.org/10.1007/978-3-319-64185-0_7
e-ISSN
16113349
ISSN
03029743
First Page
77
Last Page
89
Volume
10431 LNCS
Recommended Citation
Ye, Jingyu and Shi, Yun Qing, "A hybrid feature model for seam carving detection" (2017). Faculty Publications. 9931.
https://digitalcommons.njit.edu/fac_pubs/9931
