Steganographic Image Compression Method with Adaptive Enhancement Code Embedding and Its Performance Evaluation on 4K Standard Test Images

URI http://harp.lib.hiroshima-u.ac.jp/it-hiroshima/metadata/12391
File
Title
Steganographic Image Compression Method with Adaptive Enhancement Code Embedding and Its Performance Evaluation on 4K Standard Test Images
Author
氏名 フルカワ イサオ
ヨミ Isao FURUKAWA
別名
氏名 スズキ ジュンジ
ヨミ Junji SUZUKI
別名
Subject
steganography, image enhancement, JPEG, entropy coding, HVS, 4K image
Abstract

We have studied a steganographic image compression algorithm that embeds special codes (i.e., enhancement codes) in the encoder, and extracts these codes in the decoder to reconstruct full-resolution image. The compressed codestream from the encoder also can be decoded by standard JPEG decoder even though the resolution of reconstructed image decreases by half, so that this encoder has backward compatibility with JPEG. In embedding enhancement codes into lower DCT coefficients, there is an upper limit for the amount of codes (i.e., payload) in order to make the distortion by bit-embedding as small as possible. Although the conventional algorithm employs the constant number of scans of higher DCT coefficients to create the enhancement codes, it appears to be a bottleneck for DCT blocks that allow enough room in the payload. We propose an adaptive determination method of the enhancement codes based on the payload of each DCT block while conforming to the specifications of JPEG. This method effectively utilizes the payload in all the DCT blocks, so the quality improvement of the reconstructed image is expected. Moreover, we propose an HVS-based order rather than zigzag scan order for embedding the enhancement codes in hope of alleviating the distortion due to bit-replacement. As a result of evaluation with four 4K standard test images, the adaptive method can increase the amount of enhancement codes, and this leads to SNR improvement in comparison to the conventional constant method. Some decoded examples of both 4K/2K images are also shown.

Journal Title
広島工業大学紀要. 研究編
Volume
55
Spage
63
Epage
76
Published Date
21-Feb
Publisher
広島工業大学
ISSN
1346-9975
NCID
AA11599110
Language
eng
NIIType
Departmental Bulletin Paper
Text Version
出版社版
Set
it-hiroshima