Elimination of artifacts in encrypted binary images by modified digital halftoning techniques

An anti-counterfeiting feature, latent image, has been widely applied for banknotes and security documents. The hidden pattern of denomination by intaglio printing process can be observed by viewing the bill at certain angles of elevation. While designing the latent image, a continuous-tone cover image and a binary figurative pattern are used. The continuous tone image is halftoned by horizontal and vertical line screen, respectively. The binary figurative pattern then serves as a mask to render the corresponding area of horizontal and vertical line screen on the figurative region and background region, respectively. These procedures can be done by many available commercial softwares. However, there are various artifacts such as gaps in the junctions of horizontal and vertical screen lines, discontinuous screen lines and the white or black artifacts on the edges of the latent image. The retouching of the resulting latent image needs to be carefully and skillfully handled. In this research, we developed an automatic process to generate the artifact-free latent image inside a cover image by modified digital halftoning techniques. The methods we applied include: (1) To design new 8x8 threshold matrices in order to make a perfect joint of the horizontal and vertical screen lines. (2) To use the linear scaling adjustment to enhance the cover image not resulting discontinuous line. (3) To register the 8x8-based figurative pattern to the 8x8 threshold matrix and to avoid the visual artifacts. For latent image detection, a frequency domain treatment by FFT (Fast Fourier Transformation) and inverse-FFT is used to extract the encrypted image. This is especially useful for machine-readable applications. The results show that the developed process in this research does has the ability to automatically generate the desired latent image without any artifact. It also saves the costly retouching in the existing process. A frequency domain detection method is applied to extract latent images. The proposed techniques in this research also have great potential to proceed security printing in a digital way.