Artificial bee colony based perceptually tuned blind color image watermarking in hybrid LWT-DCT domain

Abstract

A perceptually tuned blind digital watermarking method for color images in a hybrid lifting wavelet transform (LWT) and discrete cosine transform (DCT) domain is proposed in this work. The color watermark is scrambled by arnold transform before embedding to enhance the security level. The color host image is firstly transformed by lifting wavelet transform and the approximate sub-band is partitioned into different blocks each of size 8 x 8. The pixel values of each block are quantized using a quantization table. Each quantized-block is further decomposed by the DCT, to generate the AC coefficient matrix. Scrambled watermark is thereafter inserted into the AC coefficient matrix using the perceptually tuned dynamic embedding strength factor. The visual quality and robustness of the proposed watermarking are balanced by optimizing the value of perceptually tuned strength factor by the Artificial bee colony (ABC) optimization algorithm. Experimental results show that the color watermark is undetectable (with CPSNR > 40 dB) into the host image and robust enough (with NC > 0.9) to resist the image processing and manipulation attacks. Comparison analysis with other related watermarking methods highlights the efficacy of the proposed work.