Performance Evaluation of Nested Watermarked Scheme using Objective Image Quality Metrics

Abstract

In this digital era, extensive usage of the internet has resulted in images being subjected to various distortions during processing and transmitting over open networks. For watermarking applications, watermarked image quality is evaluated in comparison to the original image. In the last decade literature, many performance metrics have been used, comprising of both subjective and objective quality metrics. The subjective image quality metrics are generally time-consuming and expensive in contrast to objective image quality metrics. This paper attempted an investigation for measuring the performance of nested watermarking schemes and objective image quality metrics which included Peak Signal to Noise Ratio (PNSR), Mean Square Error (MSE), Correlation Coefficient (CC), Normalized Correlation Coefficient (NCC), and Structural Similarity Index (SSIM). For the proposed nested watermarking method, these objective image quality metrics were evaluated using a combination of Discrete Wavelet Transforms (DWT) and Singular values (SVD) in the four frequency bands. Performance of objective metrics can be judged based on their sensitiveness to watermarking artifacts. It is clear from the results obtained, for most watermarking applications SSIM provides better insight into the performance of the algorithm used whereas PNSR fails for certain watermark artifacts. The findings drawn in this paper are based on the experimentation of a nested watermarking algorithm using a DICOM image database.