Compensating Propagated Error in Wide Angle Intra Prediction for Enhanced VVC Double Compression Detection

The distinct coding mechanisms of Versatile Video Coding (VVC) amplify challenges in Double Compression Detection (DCD). Targeting this underexplored problem, we propose a novel DCD algorithm, advancing an intra-coding analysis of the VVC error propagation characteristics within Wide Angle Intra Prediction (WAIP). In this paper, an Intra Prediction Angular Mode Retrieving (IPAMR) strategy is developed to compensate propagated error and retrieve native directional properties within intra prediction modes. Building upon the IPAMR results, a Propagated-Error-Free Prediction Mode Variation (PEF-PMV) feature is further proposed, which consists of simplified difference distribution and fluctuation intensity of prediction modes. The PEF-PMV feature effectively represents variations of intra prediction properties during multiple VVC compressions for DCD. Experimental results show that the proposed algorithm outperforms the SOTA DCD methods over the VVC standard under different coding parameters. Furthermore, ablation study verifies effectiveness of the proposed IPAMR and feature fusion strategy in VVC DCD.