Image Quality Estimation Using Logarithmic Spread Transform Dither Modulation

Abstract

As a pragmatic and novel application of digital watermarking, image quality estimation has been studied in recent years. In this paper, we propose a watermarking-based image quality evaluation scheme. The Logarithmic Spread Transform Dither Modulation is proposed based on the quantization index modulation and then applied to embed and extract the watermarks. The traditional objective metrics are employed to measure quality of images. We calculate the True Detection Rates (TDR) value to represent degradation of watermark. Considering that the embedded watermark and the watermarked image are distorted simultaneously, the image quality can be evaluated by matching the TDR value with a quality value on a pre-generated curve. Experimental results indicate that the proposed scheme provides a good image quality estimation result. The accuracy of the estimation keeps stabilization under different tested attacks, including JPEG compression, Gaussian noise addition and low-pass filtering.