Problems in halftone imaging: Synthesis and reconstruction
Abstract
A halftone image is needed when the output device is not capable of reproducing the continuous-tone image. There are two fundamental problems in halftone imaging, i.e. synthesis and reconstruction, which we will address in this work. In synthesis, the objective is coverting the continuous-tone image to a binary image in such a way that the rendered image gives the viewer the impression of tonal graduation similar to the original continuous-tone image. In this work, we will consider synthesis process for the discrete-parameter images. The objective in the reconstruction problem, which can be considered as the an inverse process to the synthesis, is reconstruction of the continuous-tone image from its halftone version. In the first part of this work, after careful study of screening and Fourier spectrum of discrete-parameter halftone image, we propose two iterative methods to reconstruct a continuous-tone image from its halftone version which has been generated by screening. To be able to apply these methods, we need to know the screen function which can be considered as the distortion operator. Thus, we estimate the screen function from the halftone image. Successive approximation is used as the first method of the reconstruction and Projection Onto Convex Sets (POCS) is used as the second method. In both methods, the optimum relaxation parameters are found and employed to speed up the convergence rate. A new technique for relaxing the constraint on the bandwidth of frequency domain operator is proposed to achieve a sharp reconstructed image. The second part of this work deals with the synthesis or design of optimum halftone image for the display device and printer. A model based approach, where the characteristics of the output device is an intrinsic part of halftoning algorithm, is introduced. In addition to the characteristics of the output device, the human visual system along with other perceptual factors are explicitly incorporated to define an error metric in order to measure the quality of the halftone image. Direct Binary Search (DBS) is used to find the halftone image which has the minimum difference between perceived, displayed/printed continuous-tone image and the perceived, displayed/printed halftone image. Since the algorithm is iterative, it is computationally intensive. In this work, we propose a lookup-based efficient method, which reduces the computational cost substantially. Although there are similarities between a display and a printer as the output devices, the differences like nonlinearity in the phosphor luminance in the display device and non-additivity caused by dot overlap for the printer, would require different approaches for modeling these devices. The spot model for the display device and printed dot for the printer are measured. The synthesis algorithm includes the gamma compensation for the display device and the effect of non-additive dot overlap for the printer.
Degree
Ph.D.
Advisors
Allebach, Purdue University.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.