Four low-complexity color trapping algorithms

Haiyin Wang, Purdue University

Abstract

Color separations (most often cyan, magenta, yellow, and black) are commonly used in printing to reproduce multi-color images. For mechanical reasons, these color separations are generally not perfectly aligned with respect to each other when they are rendered by their respective imaging stations. This phenomenon, called color plane misregistration, causes gap and halo artifacts in the printed image. Color trapping is an image processing technique that aims to reduce these artifacts by modifying the susceptible edge boundaries to create small, unnoticeable overlaps between the color planes (either at the page description language level or the rasterized image level). In this dissertation, we propose four low complexity algorithms for automatic color trapping at the rasterized image level which hide the effects of small color plane misregistrations. Our proposed algorithms are designed for software or embedded firmware implementation. The trapping method they follow is based on a hardware-friendly technique proposed by J. Trask (JTHBCT03) which is too computationally expensive for software or firmware implementation. The first two algorithms are based on the use of look-up tables (LUTs). The first LUT-based algorithm corrects all registration errors of one pixel in extent and reduces several cases of misregistration errors of two pixels in extent using only 727 Kbytes of storage space. This algorithm is particularly attractive for implementation in the embedded firmware of low-cost formatter-based printers. The second LUT-based algorithm corrects all types of misregistration errors of up to two pixels in extent using 3.7 Mbytes of storage space. This algorithm is more suitable for software implementation on host-based printers. The third algorithm is a hybrid one that combines look-up tables and feature extraction to minimize the storage requirements (724 Kbytes) while still correcting all misregistration errors of up to two pixels in extent. This algorithm is suitable for both embedded firmware implementation on low-cost formatter-based printers and software implementation on host-based printers. The fourth algorithm is developed based on the third algorithm, making use of the edge continuity and edge pairing features for further optimization and achieves the further running time reduction. Its low complexitymakes it a feasible solution for both formatter-based firmware and software based applications. All four of our proposed algorithms run, in average, more than three times faster than a software implementation of JTHBCT03.

Degree

Ph.D.

Advisors

Allebach, Purdue University.

Subject Area

Electrical engineering

COinS