A Perceptually Adaptive Scheme for Image Bit-Inversion-Based Watermarks

Abstract

A perceptually adaptive scheme for image watermarking by bit inverting transformation is developed. The transformation performs the change of a signal level by both inverting a bit and changing the level at random in a limited range. From the property analysis of the transformation, a perceptual model of two metrics of level distortion is assumed for evaluating a subjective visual quality of the transformed image. The measurements of image quality by the human subjective evaluation are analyzed and then, the evaluation function that can yield an estimate of the subjective visual quality from the two distortion measurements of the transformed image is determined. By using the evaluation function, every image region is transformed with its parameter adaptively changed so as to achieve a desired visual quality. The simulation results demonstrate the performance of the evaluation function and show a linear correlation between the evaluated values and the measurements of visual quality.