Performance Analysis of Watermarking using SVD of watermark in Non-sinusoidal Column and Row Transforms

A novel watermarking technique using Singular Value Decomposition (SVD) and non-sinusoidal column/row transforms like Haar, Walsh, Slant and Discrete Kekre Transform is proposed in the paper. Host images are subject to column/row transform using orthogonal non-sinusoidal transforms and watermark is subjected to SVD. To prevent loss of watermark after performing attacks on watermarked image, watermark is embedded into mid-frequency band of host image. Singular values of watermark are inserted in the mid frequency band. Performance of proposed technique is observed against following attacks: cropping, compression (using transforms, JPEG compression and Vector Quantization), resizing (using transforms, grid based interpolation and bicubic interpolation) and noise addition. Robustness of proposed technique is measured using Mean Absolute Error (MAE) between embedded watermark and the one recovered from attack. Overall performance of proposed technique is robust against transform based resizing and binary and Gaussian distributed run length noise addition attack. KeywordsWatermarking;, column & row transforms; Singular Value Decomposition; Haar; Slant; DKT; Walsh African Journal of Computing & ICT Reference Format: H.B Kekre, T. Sarode & S. Natu (2014). Performance Analysis of Watermarking using SVD of watermark in Nonsinusoidal Column and Row Transforms. Afr J. of Comp & ICTs. Vol 7, No. 5. Pp9-22.