Abstract

Abstract:- Symbols are watermarked into various images, marketing documents, videos and multimedia files before circulating on social media for authenticity and ownership. Digital image watermarking techniques is found suitable for invisible symbol watermarking that is imperceptible and robust to common attacks. In this paper, an image watermarking technique using adaptive wavelet texture features domain through encryption and Bhattacharya distance. The fundamental concept explored is to fragment the carrier image in time domain into four partial non overlapping regions whereas symbol image is encrypted. Symbol image is encrypted based on Rubik’s Cube principle through random key generated. Subsequently statistical parameters such as mean (degree 1), standard deviation (degree 2), skew (degree 3), kurtosis (degree 4) and variance are calculated. These parameters are calculated through wavelet coefficients obtained from the regions of carrier image and encrypted symbol, which are further applied to obtain Bhattacharya distance. Minimum Bhattacharya distance is applied criteria between the subbands of regions of carrier image and symbol for embedding in wavelet domain. Also the embedding coefficient is calculated through Bhattacharya distance for invisible watermarking. Robustness, imperceptibility and signal to noise ratio were measured on multiple carrier and symbol images under common attacks, which yield better results. Keywords: Adaptive watermarking technique, Encryption, Texture feature, Wavelet Transform, Skew, Kurtosis, Variance, Bhattacharya distance, Image processing attacks