PROJECT TITLE:

Towards Practical Self-Embedding for JPEG-Compressed Digital Images - 2015

ABSTRACT:

This paper deals with the planning of a practical self-recovery mechanism for lossy compressed JPEG pictures. We have a tendency to extend a recently proposed model of the content reconstruction problem based on digital fountain codes to require into consideration the impact of emerging watermark extraction and block classification errors. In distinction to existing strategies, our theme guarantees a high and stable level of reconstruction quality. Instead of introducing reconstruction artifacts, emerging watermark extraction errors penalize the achievable tampering rates. We introduce new mechanisms that enable for handling high-resolution and color pictures efficiently. In order to analyze the behavior of our theme, we have a tendency to derive an improved model to calculate the reconstruction success probability. We introduce a brand new hybrid mechanism for spreading the reference information over the whole image, which permits to search out a good balance between the achievable tampering rates and therefore the computational complexity. Such an approach reduced the watermark embedding time from the order of many minutes to the order of single seconds, even on mobile devices.