Safe-LBP: A visually meaningful image encryption scheme based on LBP and compressive sensing

Abstract

Recently, some visually meaningful image encryption schemes have been proposed to protect the visual security of images. However, much of the existing literature on visual image encryption ignores the features of cover images. To address the above issues, this paper proposes a content-adaptive visual image encryption scheme based on piecewise linear chaotic map (PWLCM), compressive sensing (CS), and safe local binary pattern (Safe-LBP). First of all, the plain image are sparse by applying discrete wavelet transform (DWT). Secondly, the cipher data is obtained by encrypting and compressing the sparse matrix using Arnold scrambling and CS. Next, we design a new LBP operator to discover appropriate embedding place in the cover image, which tremendously improves the imperceptibility of the cipher image. Finally, the visually meaningful steganographic image is obtained by embedded all cipher data through the least significant bit (LSB) method. Simulations and comparisons show that the proposed algorithm has excellent visual security and decryption quality and proves its robustness against various types of noises and data loss.