Signal and Image Processing Seminar

Digital halftoning is the process by which a continuous-tone image is converted to a binary image, or halftone, for printing or display on binary devices. Error diffusion is a halftoning method which employs feedback to preserve the local image intensity and reduce low frequency quantization noise. It is a highly nonlinear process, and it is therefore difficult to analyze mathematically. In this work, we present a linear gain model for the quantizer which accurately predicts the edge sharpening and noise shaping effects of error diffusion. We use the model to construct a residual image that has a low correlation with the original image. By weighting this residual with a model of the human visual system, we obtain a measure of the subjective effect of the quantization noise on the viewer. We further obtain a distortion metric for the halftoning scheme. By characterizing the edge sharpening, noise shaping, and distortion of an error diffusion scheme, we obtain objective measures of subjective quality of halftones. This permits the comparison of halftoning schemes.

We present a new, efficient inverse halftoning scheme for error diffused halftones that produces results comparable to the best current methods, but at a fraction of the computational cost. We demonstrate a method of modeling inverse halftoning schemes, and use the model to generate residual images, which we weight with the human visual system model. We also compute an effective transfer function for the inverse halftoning scheme. By characterizing the degree of blurring and the noise content, we obtain objective measures of subjective quality of inverse halftones. This allows competing inverse halftoning algorithms to be compared. We further use the linear gain model to design and analyze the performance of applications which include error diffusion. We again make use of the model of the human visual system to obtain objective measures of the quality of images produced by these applications.