Adaptive Image Steganography Based on Edge Detection Over Dual-Tree Complex Wavelet Transform

The proposed method aims for an advanced steganographic approach based on an adaptive embedding process using edge detection over Dual Tree Complex Wavelet Transform (DT-CWT). Here, subband coefficients allow for maintaining high image imperceptibility even with a dense embedding of secret data. Prior to the embedding process, the cover image is divided into multiple non-overlapping blocks and the secret data bits are indirectly concealed in the selected subbands of DT-CWT coefficients. Amount of data bits embedded on different patches depends on the high frequency elements in each patch. These high frequency regions are identified by using Canny edge detection technique. This helps to embed more bits over highly textured regions and fewer bits over smooth regions and hence significantly reduce the distortion of the stego-image. The DT-CWT provides multiple subbands along multiple orientations increasing data capacity with high cover-stego image and secret-recovered image PSNR value. The performance is evaluated on the basis of different standard benchmarks like similarity index, PSNR, payload capacity etc. to evaluate different aspects of image steganography.