The performance of Circular and Adaptive Embedding Methods in Imperceptible Data Hiding using DWT

Abstract

This work investigates three watermark embedding schemes in the discrete wavelet transform (DWT) domain: circular, adaptive circular, and texture-adaptive circular embedding. Each method embeds a binary watermark into the LL subband of a 1-level DWT using concentric rings to define the spatial pattern. Circular embedding applies uniform strength across rings, adaptive circular embedding adjusts inner-ring strength to account for human visual perception, and texture-adaptive circular embedding further modulates strength based on local image variance. The paper presents a fair and normalized comparison of imperceptibility among the three methods using PSNR and SSIM metrics.