Accelerating Encrypted Watermarking using Wavelet Transform and CKKS Homomorphic Encryption

Abstract

Encrypted watermarking enables secret watermark embedding on a public platform such as blockchain for better transparency. It offers data and computation traceability to ensure a proveable watermark embedding and validation. Nevertheless, encrypted watermarking has a setback regarding high computation cost and long computation time that hinder its implementation on most blockchain platforms. In this paper, we propose a joint use of discrete wavelet transform (DWT) and CKKS (Cheon-Kim-Kim-Song) homomorphic encryption to speed up and improve the efficiency of encrypted watermarking. The DWT reduces the size of encrypted data to 2L where L refers to the DWT level. Meanwhile, the CKKS encryption speed up the encrypted computation using approximate arithmetic computation and predefined precision. The result shows that CKKS and DWT level 2 is the most optimal solution delivering up to ≈ 27.24 × faster computation and ≈ 5.48 × lesser memory compared to the existing method (BFV-encryption and DCT). In addition, the proposed method has a similar watermarking quality and watermark extractability to non-encrypted watermarking.