Susan Farnand

Abstract

Imaging system development often involves impact assessment of design choices. For systems that generate images for human consumption, such as cameras and displays, the effect of design decisions are often evaluated using `real-world' images. System changes can have complicated effects on pictorial images that do not, as yet, have specified instrumental measurement methods. Consequently, human observers are often used in image quality assessment. However, human observers can react differently to complex pictorial stimuli both between observers and for a single observer over the course of a lengthy experiment. In an experimental setting, pictorial scenes present a greater opportunity than do uniform patches for observers' individual differences to significantly impact the process. This study was conducted to increase the understanding of the optimal design of pictorial stimuli for more effective and efficient perceptual experiments. The goals of this dissertation were to: 1. Understand the impact of image content on visual attention and the consistency of image comparison experimental results 2. Understand how visual attention changes with successive viewing of pictorial images 3. Apply this understanding to develop guidelines for pictorial target design for perceptual image comparison experiments To achieve these objectives, a series of experiments were conducted to evaluate the impact of pictorial scene complexity on fixation and experimental response consistency. For these experiments, scenes exhibiting a range of perceived complexity were required. To select appropriate scenes, the concept of what constitutes a complex image was first considered. Experiment I was conducted to evaluate the number of areas perceived to be important in a variety of scenes. Observers were asked to identify the important areas of pictorial scenes. The scenes were also electronically segmented. The results from Experiment I were used to select scenes that provided a range of complexity for stimuli in Experiment II. This test examined the impact of image complexity on observer viewing behavior. Along with evaluating eye movements, observers were asked to describe the test scenes using up to five keywords. The results of Experiments I & II indicate that perceptual methods, segmentation, and eye-tracking generally provided consistent results with regard to image complexity. The exceptions involved issues of scale such that scenes viewed from afar blended into one significant object while one object viewed up close lacked a point of focus. The results of Experiment II were used to generate a proposal for guidelines for designing pictorial stimuli for image comparison experiments. Using these guidelines, scenes were selected and tested in Experiment III. The fixation consistency results of this experiment were generally as expected. However, fixation consistency did not always equate to experimental response consistency. Along with scene complexity, the image modifications (global versus local) and the difficulty of making the image equivalency decisions played a role in the experimental response as well. The results of Experiment III were used to confirm and augment the proposed guidelines. The guidelines developed in this study will benefit those conducting perceptual experiments with pictorial stimuli. Specific examples include color reproduction, perceptual color standards, and image equivalency research. A better understanding of what makes images equivalent may be useful in developing automated approaches to measuring image quality. And the guidelines may be useful in the improvement of the quality of images themselves. Fredembach (2011) has proposed that perceived image quality can be improved by increasing the perceived saliency of the main subject matter. These guidelines, including the use of blur, will be helpful in achieving this aim.

Library of Congress Subject Headings

Visual perception--Testing; Imaging systems--Image quality--Research; Attention--Research

Publication Date

6-10-2013

Document Type

Dissertation

Student Type

Graduate

Department, Program, or Center

Chester F. Carlson Center for Imaging Science (COS)

Advisor

Bailey, Margaret

Advisor/Committee Member

Fairchild, Mark

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: BF241 .F37 2013

Campus

RIT – Main Campus

Plan Codes

CLRS-PHD

Download

Share

COinS