Skip to Main Content
Digital image watermarking algorithms embed identifying marks that can be used for authentication. To make the distortions induced by the embedding process imperceptible, the watermarking algorithm must determine their visual threshold. This paper presents: (1) the results of a psychophysical detection experiment for wavelet-distortion targets presented against textured backgrounds of varying contrasts; and (2) an associated wavelet-based image watermarking algorithm that embeds the watermark at the threshold of visibility by weighting it using a local masking measure prior to embedding. The embedded watermark is shown to be robust to JPEG compression at quality factors as low as 20.