Texture control in digital halftoning

There are two ways to represent visual information: photographic and artistic. Photographic approaches attempt to approximate an image despite the limitations of the output medium. Traditional halftoning takes a photorealistic approach. In an artistic rendering visual information is interpreted by the artist and displayed accordingly using the chosen medium. The non-photorealistic rendering area of computer graphics develops tools and techniques to enable interpretive rendering in digital media.

Texture is an inevitable artifact of halftoning. The challenge of photorealistic halftoning is to preserve image features—tone, edges, and textures—and to hide the halftoning texture. To the contrary, in artistic rendering texture is often used as a visual cue and an expressive mean. In this thesis I explore the use of halftoning texture to enhance the representation of visual information in both photorealistic and interpretive rendering applications. However, this use of texture requires methods and techniques to control texture in halftoned images. Thus, the objective of this work is to control the appearance of texture in the resulting images.

My technique of texture control is based on previous halftoning algorithms: ordered dither and error diffusion. I use the ability of the ordered dither algorithm to define halftoning texture through the arrangement of threshold values in its dither matrix. The thesis describes two methods of generating dither matrices: image processing and procedural texturing. The use of texture based dither matrices guarantees the appearance of desired textures in the halftoned image. The strength of the resulting texture is controlled by combining ordered dithering with the error diffusion process.

The ability to define and control texture in the halftoned image leads to the use of this texture as an expressive mean in image rendering. A user may introduce a variety of artistic effects into the image. Examples include embossing an image with a texture or text; approximation of traditional art styles and rendering techniques—pencil drawing, carving, oil brush painting. The thesis also includes techniques that allow us to map texture features to enhance representation of image gradient, 3-D scene information and subjective user defined information.

This study is a contribution to both photorealistic and artistic halftoning of images. It is a new approach to non-photorealistic rendering. Unlike previous techniques, interpretive halftoning is not limited to any particular style of rendering. Moreover, artistic effects generated by previous techniques may be approximated.

Investigation of texture control in halftoning extends photorealistic dithering techniques. It turns out that the use of image based dither matrix improves rendering of the original image textures and edges. Also, the thesis includes investigation of an image quality measure that allows us to analyze halftoned images. This measure is based on multi-scale analysis of image edges and thus enables us to quantify edge distortions introduced by the halftoning algorithm.