ハイブリッドウェーブレット変換、K平均特異値分解、および空間記憶誘導型猫群最適化技術による透かし埋め込み

Integration and use of multimedia data has increased primarily due to advancements in technology and widespread utilization of internet services, especially with respect to multimedia audio, images, and videos. Such rapid progression in technology has highlighted the need for effective intellectual property and copyright protection. In response to these issues, this paper discusses the development of an improved watermarking technique for the embedding of watermarks in compressed videos. The newly developed technique uses Discrete Wavelet Transform (DWT) in conjunction with K-Means Singular Value Decomposition (K-SVD) to facilitate precision watermarking with marginal loss of quality. The optimization technique is revised to incorporate a Spatial Memory Guided Cat Swarm Optimization (SMGCSO) technique, which mimics the hunting behavior of cats to avoid local optima trap. Three core phases constitute SMGCSO: an exploration phase to seek optimal new positions, a spatial memory phase where best positions are stored, and an exploitation phase to refine the best solution, thus ensuring the global maximum is achieved. In terms of watermarking, Peak Signal-to-Noise Ratio (PSNR) and Bit Error Rate (BER) determine performance. Watermarking, video processing attacks, and video copyright protection illustrate the practical utilization of the DWT-K-SVD-SMGCSO technique. There is an approximate 28% increase in PSNR and 45% decline in BER relative to the baseline DWT-K-SVD technique. Such advancements demonstrate superior resistance to video processing attacks and enhanced watermark robustness.