LSBs-based quantum color images watermarking algorithm in edge region

Abstract

Based on the NEQR representation for quantum color and binary images, an enhanced quantum watermarking scheme is investigated through Gray code transform and least significant bit (LSB) steganography, which embeds a quantum binary image (i.e., watermark image) into the edge region of a quantum color image (i.e., carrier image) LSB and second LSB. The size of the carrier and watermark images are assumed to be $$ 2^{n} \times 2^{n} $$ and $$ 2^{n - 1} \times 2^{n - 1} $$, respectively. At first, the watermark image is resized into an appropriate size image with 4-qubit grayscale based on the nearest neighbor interpolation method, which is of the same size with the preselected edge region in carrier image. To enhance the security of the watermark image, the binary code of 4-qubit grayscale of watermark image is transformed into the corresponding Gray code, and one 3-Controlled-NOT gate is utilized to generate a quantum binary image $$ \left| {K1} \right\rangle $$. To further scatter the watermark image qubits that are embedded into the LSB and second LSB of carrier image, the quantum image $$ \left| {K1} \right\rangle $$ is employed to choose any two channels from the color image among the three channels of R, G and B (i.e., R, G or R, B channels would be chosen as the embedding channels). Furthermore, a quantum binary image $$ \left| {K2} \right\rangle $$ is generated through XOR operation decided by the quantum image $$ \left| {K1} \right\rangle $$, which is used to determine the embedding order of watermark image qubits. The extraction process is the inverse operation of embedding, which also needs the two quantum binary key images $$ \left| {K1} \right\rangle $$ and $$ \left| {K2} \right\rangle $$. Finally, the experiment results are simulated under the classical computer software MATLAB 2016(b), which illustrates that our investigated LSBs-based quantum watermarking has a better visual effect than some related works in terms of PSNR value.