Enhancing robustness in video data hiding against recompression with a wide parameter range

Abstract

With the exponential growth of online videos and the availability of advanced video editing tools, the integrity of videos is increasingly threatened by various cyber-based attacks, particularly recompression attacks. Different recompression parameters are used for various attack scenarios. However, recompression operations can modify video data and cause authentication failures, especially for videos with integrated authentication information. To address this issue, this research proposes a robust video data concealing strategy to protect authentication information during recompression operations. By analyzing the impact of recompression on the reference and encoded components of a video, a robust embedding entity is developed. This entity enables the identification of suitable embedding parameters that minimize distortion caused by recompression. Based on the embedding entity and parameters, a scheme is proposed to hide data using two embedding methods, thereby enhancing the robustness of the overall approach. The performance of the scheme is evaluated based on criteria such as the correct rate of extracted information and the quality of marked videos. Hypothesis testing is employed to assess the confidence level. In comparison to existing robust video data hiding methods against recompression, our scheme demonstrates the ability to preserve more information during recompression, particularly with Quantization Parameters (QP) ranging from 25 to 34. This is achieved when the test videos are initially compressed with QP=28.