A Visual Complexity-sensitive DWT Ordering Scheme for Hiding Data in Images

Abstract

A scheme is proposed to hide data in images based on a prioritized ordering of the content of the host (or cover) image. A watermark embedding process uses the watermark strength to determine the ordering of the 16 regions resulting from the second level Discrete Wavelet Transform (DWT) decomposed content of the host image. To determine the best ordering for hiding the data, various types of image of varying content and, hence, visual complexity were considered, analyzed and ranked in terms of their ability to withstand changes that do not imperil the visual quality (PSNR) of their watermarked versions depending on which an N×N-sized watermark stream is hidden in the 8 highest ranked sub-bands of the host image. From this perspective, an ordering for images classified as simple, normal and complex images was used to determine a generalized ordering of the DWT decomposed sub-bands of the image. The generalized ordering presented here ascertains that the content of the image and its visual complexity had little effect on an earlier proposed prioritized ordering of the DWT Sub-bands. To validate the veracity of the ordering scheme, 1000 images from the Corel 1000A database, their visual complexity and features were taken into account. The results confirmed that high embedding capacity, appreciable visual quality of watermarked images and complete recovery hidden data are realizable based on the ordering scheme proposed in this study.