Asymmetric color image encryption and compression based on discrete cosine transform in Fresnel domain

• Published in: Results in optics Vol. 1; p. 100005
• Main Authors: Kumari, Eakta, Mukherjee, Saurabh, Singh, Phool, Kumar, Ravi
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.11.2020; Elsevier
• Subjects: Discrete cosine transform; Fourier transform; Fresnel propagation; Optical encryption; Fourier transform; Fresnel propagation; Optical encryption; Discrete cosine transform
• ISSN: 2666-9501; 2666-9501
• DOI: 10.1016/j.rio.2020.100005

•New asymmetric method for color image encryption and compression is proposed.•Proposed method is robust to noise occlusion and special iterative attacks.•Comparative analysis confirms the advantages of proposed method over existing similar methods.•Storage and secure transmission of color images improved significantly without much loss of information. In this paper, a new asymmetric method is presented to encrypt and compress color images using discrete cosine transform in Fresnel domain. First, the input color image is decomposed into three channels: red, green and blue. Each channel image is converted into phase image and bonded with an amplitude mask followed by Fresnel transform to get the complex intermediate image. Phase reservation (PR) and phase truncation (PT) operation are performed on the intermediate image. Thereafter, a random phase mask is bonded with the amplitude part and the phase part serves as of the first private key for decryption. The complex image is Fresnel propagated followed by the second PR and PT operation. Phase part serve as another private key and the discrete cosine transform is applied on the amplitude part of each red, green and blue channel to get the corresponding compressed encrypted images. The final compressed color encrypted image is obtained by combining the encrypted images of all three channels. In the decryption process, encrypted image is decomposed in red, green and blue channel and lost data in each channel is replaced by zeros. Each channel is subject to inverse discrete cosine transform and cascaded PT/PR operations with Fresnel propagation in reverse directions and recombined to get the original image. Numerical simulation results are presented to validate the proposed method. Also, the robustness of the proposed method is checked against various existing attacks and a comparative analysis confirms the effectiveness of our method.