IMAGE HALF-TONING WITH CELLULAR NEURAL NETWORKS

Many algorithms have been devised for halftoning digital images. These algorithms all suffer well-studied defects, which are especially appa rent in the case where the resulting image is displayed at the margina lly oversampled resolution and viewed at the critical pixel merge dist ance. Recently, it has been shown that a neural network approach may b e useful for halftoning. Here, the feasibility of using neural network s in a practical application is considered. The cellular neural networ k (CNN) architecture is chosen for its proven implementability in VLSI and high speed operation. Since both the CNN and halftoning have a ge ometrically local character, the CNN provides a natural implementation . The CNN template weights are derived by analogy to the well-known er ror diffusion algorithm for halftoning. Some limitations of the neural network approach are analyzed providing an advance in designing templ ate weights over previous methods. These limitations are shown to be e specially critical in the case of the small interconnection neighborho ods needed for efficient implementation. Our design criteria are valid ated by direct simulation. The resulting halftones are shown to be mor e faithful reproductions of the original than those produced by the er ror diffusion algorithm. It is suggested that a CNN with optical input s could provide a high-speed scanner/halftoner for applications such a s FAX.