Optical image encryption and information hiding with computational ghost imaging based on a forgery attack mechanism

Computational ghost imaging (CGI) has shown considerable potential for optical encryption and information hiding. However, recent forgery attacks reveal that CGI reconstruction can be manipulated by replacing the original masks with forged ones, while this controllability has mainly been regarded as a security vulnerability. Here, this characteristic is exploited as an authorized information-hiding mechanism. Secret bits are embedded into visually similar target images using pairwise differential block coding with Hann-window weighting, and the corresponding forged masks are obtained through inverse optimization without altering the acquired single-pixel measurements. Hidden information is recovered from the reference and information-bearing reconstructions by matched-filter differential decoding and multi-image decision fusion. Statistical analysis, numerical simulations, and optical experiments demonstrate controllable reconstruction and reliable hidden-information recovery. This work extends forgery-based reconstruction controllability from attack analysis to authorized information hiding, providing a potential approach for CGI-based optical security.