Embedding Information into Printed Matter Using Invisible Luminance-modulated Illumination

This paper presents a novel technique for embedding information in printed images of real objects captured by a camera, aiming to verify the authenticity of these images. The key feature of this method is that information is invisibly embedded in the illumination light that shines on the object when capturing image of the real object. As a result, the captured image of the object, as well as the printed version of that image, retains this invisible information. In this study, we focus on face photographs, commonly used for ID photos. A projector, serving as the light source, projects an image with uniform brightness across the illumination area. To embed 1 bit of information in each 16x16 pixel block, a small value is assigned to a specific frequency component based on binary data. The printed image is scanned, re-digitized, and subjected to DCT transformation, allowing information to be extracted from the values of specific frequency components. Experimental results demonstrated that 100 % accurate information embedding and retrieval are achievable with appropriate parameter settings. This technique represents a cost-effective and practical solution for ensuring the authenticity of printed ID images, offering significant advantages over existing methods by utilizing standard consumer-grade devices. The findings could be instrumental in preventing identity fraud and expanding applications in security and authentication.