We intend to largely disseminate the main results of RDH-COMPRESS project. Thus, we expect to publish our results (high capacity RDH embedding, RDH-compressors) in 3-4 journals with high impact in image processing (IEEE Trans. on Image Processing, IEEE Trans. on Multimedia, IEEE Trans. Inf. Forensics and Security) as well as in 5-6 papers in selective conferences (IEEE Intl. Conf on Image Process (ICIP), IEEE Int. Conf. on Multimedia and Expo (ICME), IEEE Intl. Conf. on Acoustics, Speech, and Signal Process., etc.). Dissemination will be also provided via the site of the project.
Albu F., Set-membership Sparsity-Aware Proportionate Normalized Least Mean Square Algorithms for Active Noise Control, 2020
Radhika S., Albu F., Chandrasekar A., Steady State Mean Square Analysis of Standard Maximum Versoria Criterion Based Adaptive Algorithm, 2020
IEEE Signal Processing Letters
On Preserving Histogram Aspect in Prediction Error Expansion RDH
Authors: Dinu Coltuc, Dragoi Ioan-Catalin, Coanda Henri-George
This paper proposes a modified PEE RDH scheme that preserves the natural aspect of the prediction
error histogram. Compared with the standard PEE schemes, there are two differences. The addition of the
data bit to the expanded prediction error is replaced by a new procedure that considers either addition
or subtraction. The switch between addition and subtraction is controlled by the predicted pixel value.
A new bitrate control scheme is introduced. Instead of threshold control, the threshold is fixed to a
rather large value and the bitrate is controlled by using the context smoothness. Experimental results
are presented
IEEE Signal Processing Letters
Low Distortion LSB Replacement
Authors: Dinu Coltuc
This letter deals with low distortion Least Significant Bit (LSB) replacement. Pairs of bits are encoded
on 3 bits and embedded into 3 pixels such that at most one LSB is modified. One gets 2/3 bpp at an expected
number of modifications per pixel (ENMP) of 0.250. The pairwise LSB replacement is further extended to 1 bpp
by embedding groups of n bits into n + 1 pixels. Closed form expressions are given for the ENMP.
The fine tuning of the bitrate in the range (2/3,1) bpp is also discussed. Theoretically, up to 2/3 bpp,
the proposed LSB replacement operates at the same distortion as the state-of-theart LSB matching of Mielikainen.
Practically, depending on the percentage of saturated pixels, the proposed LSB replacement can outperform the
low distortion state-of-the-art LSB matching on a larger range of bitrates.
Ionita Marius-Giorgian, Coanda Henri-George, An improved automatic periodic noise removal algorithm for microscopic images, 2022
ECAI Conference Program
