Share Email Print
cover

Spie Press Book

Computational Image Quality
Author(s): Ruud Janssen
Format Member Price Non-Member Price
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Book Description

Images are a powerful, efficient means for communicating information. This book looks at metrics and methods for predicting image quality based on human visual and cognitive information-processing capabilities.

Book Details

Date Published: 25 June 2001
Pages: 158
ISBN: 9780819441324
Volume: PM101

Table of Contents
SHOW Table of Contents | HIDE Table of Contents
1 Introduction 1
2 Approaches to Image Quality 7
2.1 Statistical measures 7
2.2 Measures using visual front-end models 9
2.3 Measures incorporating modulation transfer functions 10
2.4 Multidimensional impairment measures 12
2.5 Task performance measures 14
2.5.1 Diagnostic accuracy 15
2.5.2 Reconnaissance 16
2.5.3 Text readability 16
2.6 Measures for color reproduction quality 16
2.7 The approach presented in this book 18
3 Image Quality Semantics 19
3.1 Introduction . 20
3.2 Quality and information processing 21
3.2.1 Understanding information-processing systems. 21
3.2.2 The quality of information 21
3.2.3 The quality of images 22
3.3 Image quality semantics 22
3.3.1 Image processing by the visuo-cognitive system 22
3.3.2 Naturalness, usefulness, and quality 23
3.4 Experiments . 25
3.4.1 Experiment 1: Influences of naturalness and
usefulness on image quality 25
3.4.2 Experiment 2: Image quality regarded as a
compromise between naturalness and usefulness 31
3.5 Concluding remarks 36
4 Visual Metrics: Discriminative Power through Flexibility 39
4.1 Introduction . 39
4.2 The usefulness of flexibility 42
4.3 Recipe for an optimal metric 46
4.3.1 Problem specification 46
4.3.2 Solution 50
4.3.3 Flexibility versus rigidity: performances compared 53
4.3.4 Concluding remarks 55
4.4 Vision and visual memory 57
4.4.1 Visual identification: vision versus memory 57
4.4.2 Calibrating visual metrics 57
4.4.3 Uncalibrated visual metrics: partial flexibility 59
4.5 Conclusions 61
4.6 Probability of a topological error 62
5 Predicting the Usefulness and Naturalness of Color Reproductions . 65
5.1 Introduction 66
5.2 Metrics for brightness and color 69
5.3 Predicting usefulness 74
5.3.1 Discriminability 74
5.3.2 Results and discussion 77
5.4 Predicting naturalness 82
5.4.1 The construction of memory standards 82
5.4.2 Matching perceived object colors with memory
standards85
5.4.3 Results and discussion 87
5.5 Conclusions . 92
6 Image Quality Revisited 95
6.1 Introduction . 95
6.2 What is image quality? . 96
6.3 The internal quantification of attributes 97
6.4 An optimal metric for overall discriminability 100
6.4.1 A measure for overall discriminability 101
6.4.2 Optimizing overall discriminability 102
6.5 An optimal metric for overall identifiability 104
6.5.1 The accumulation of scale value distributions 104
6.5.2 A decision rule for identification 106
6.5.3 A measure for overall identifiability 107
6.5.4 Optimizing overall identifiability 108
6.6 Estimating the number of discriminable and identifiable items 109
6.7 Partial flexibility 110
6.7.1 Discriminability for partial flexibility 113
6.7.2 Identifiability for partial flexibility 113
6.7.3 Re-estimating the number of discriminable and
identifiable items 114
6.7.4 Optimizing the degree of flexibility 117
6.8 Application: black-and-white images of natural scenes 119
6.9 Conclusions 125
7 Epilogue 131
Bibliography 137
Sample stimuli 143
Subject Index 147
Author Biography

© SPIE. Terms of Use
Back to Top