Evaluation and optimal design of spectral sensitivities for digital color imaging

Show full item record

Title: Evaluation and optimal design of spectral sensitivities for digital color imaging
Author: Quan, Shuxue
Abstract: The quality of an image captured by color imaging system primarily depends on three factors: sensor spectral sensitivity, illumination and scene. While illumination is very important to be known, the sensitivity characteristics is critical to the success of imaging applications, and is necessary to be optimally designed under practical constraints. The ultimate image quality is judged subjectively by human visual system. This dissertation addresses the evaluation and optimal design of spectral sensitivity functions for digital color imaging devices. Color imaging fundamentals and device characterization are discussed in the first place. For the evaluation of spectral sensitivity functions, this dissertation concentrates on the consideration of imaging noise characteristics. Both signal-independent and signal-dependent noises form an imaging noise model and noises will be propagated while signal is processed. A new colorimetric quality metric, unified measure of goodness (UMG), which addresses color accuracy and noise performance simultaneously, is introduced and compared with other available quality metrics. Through comparison, UMG is designated as a primary evaluation metric. On the optimal design of spectral sensitivity functions, three generic approaches, optimization through enumeration evaluation, optimization of parameterized functions, and optimization of additional channel, are analyzed in the case of the filter fabrication process is unknown. Otherwise a hierarchical design approach is introduced, which emphasizes the use of the primary metric but the initial optimization results are refined through the application of multiple secondary metrics. Finally the validity of UMG as a primary metric and the hierarchical approach are experimentally tested and verified.
Record URI: http://hdl.handle.net/1850/15020
Date: 2002-04

Files in this item

Files Size Format View
SQuanThesis04-2002.pdf 11.20Mb PDF View/Open

The following license files are associated with this item:

This item appears in the following Collection(s)

Show full item record

Search RIT DML


Advanced Search

Browse