Color imaging: Efficient transformations and model based halftoning

Ufuk Ayse Agar, Purdue University

Abstract

We investigate two problems in color imaging. The first problem is extraction of optimum grid structures that will minimize the error in the interpolation of nonlinear multidimensional color transformations. We use an efficient and fast multidimensional linear function interpolation technique called sequential linear interpolation (SLI) which uses a structured lookup table with nonuniformly spaced grid points. Our error criterion is the maximum absolute interpolation error. We develop two near-optimal minimax SLI grid structure design methods to allocate a fixed number of grid points. The first method makes use of an asymptotic theory and gives closed form expressions for the design parameters. The second method is iterative and creates grid structures using a greedy approach. Preliminary numerical simulations indicate the usefulness of these two design techniques for the problem of color printer characterization. The second problem is model based color halftoning. We propose a halftoning method using the direct binary search (DBS) algorithm. Our method strives to minimize the perceived error between the continuous tone original color image and the color halftone image. We exploit the differences in how the human viewers respond to luminance and chrominance information and use the total squared error in a luminance/chrominance based space as our metric. Our method also incorporates a measurement based color printer model to account for dot overlap and to increase the color resolution and the visual quality of halftones. We calibrate our halftoning algorithm to ensure accurate colorant distributions in resulting halftones. Our algorithm creates visually pleasing and colorantly correct halftones.

Degree

Ph.D.

Advisors

Allebach, Purdue University.

Subject Area

Electrical engineering

COinS