Video watermarking of live streamed MPEG-4 frames using ELM-Fuzzy-PSO hybrid scheme

The widespread expansion of internet and digital sharing of multimedia content has led to a huge rise in security concerns like ownership, content authentication and copyright protection of the multimedia data. This paper presents a novel scheme for watermarking of MPEG-4 videos based on the Extreme Learning Machine (ELM) for frame selection task and hybridization of Fuzzy logic with Particle Swarm Optimization technique for watermark embedding and extraction process. The ELM is applied on the host video to select the relevant frames fit for watermarking process. The embedding step is carried out in DWT-SVD domain using Fuzzy-PSO technique. The proposed scheme is tested over five standard videos having advanced video coding (AVC) format. The robustness and visual quality testing post embedding is respectively carried out by computing Normalized Cross-Correlation (NC(W,W’)) and Bit Error Rate (BER(W,W’)) along with Average Peak-Signal-to-Noise Ratio ( PSNR_avg ) parameters by applying eight different attacks. Our proposed scheme is tested and compared for its outcomes with those of different other standard schemes. It is found that our results are superior to all other frontline schemes.