koschan, abidi - digital color image processing 2008

DIGITAL COLOR IMAGE P RO C E SSIN G DIGITAL COLOR IMAGE PROCESSING Andreas Koschan Mongi Abidi WILEYINTERSCIENCE A JOHN WILEY & SONS, INC., PUBLICATION Copyright 02008 by John Wiley & Sons, Inc All rights reserved Published by John Wiley & Sons, Inc., Hoboken, New Jersey Published simultaneously in Canada N o part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 ofthe 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 11 River Street, Hoboken NJ 07030, (201) 748-601 1, fax (201) 748-6008, or online at http:liwww.wiley.com/go/perniission Limit of Liability/’Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose No warranty may be created or extended by sales representatives or written sales materials The advice and strategies contained herein may not be suitable for your situation You should consult with a professional where appropriate Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages including but not limited to special, incidental, consequential, or other damages For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax(317) 572-4002 Wiley also publishes its books in a variety of electronic formats Some content that appears in print may not be available in electronic format For information about Wiley products, visit our web site at nww.v, iley.com Library of Congress Cataloging-in-Publication Data: Koschan, Andreas, 1956Digital color image processing / b y Andreas Koschan and Mongi A Abidi p, cm ISBN 978-0-470- 14708-5 (cloth) Image processing-Digital techniques Color I Abidi Mongi A 11 Title TA1637.K678 2008 621.36’74~22 2007027369 Printed in the United States of America I to my daughter Andrea (Andreas Koschan) in memory of my father Ali (Mongi Abidi) TABLE OF CONTENTS Preface Acknowledgment Introduction Goal and Content of this Book 1.1 1.2 Terminology in Color Image Processing 1.2.1 What Is a Digital Color Image? 1.2.2 Derivative of a Color Image 1.2.3 Color Edges 1.2.4 Color Constancy 1.2.5 Contrast of a Color Image 1.2.6 Noise in Color Images 1.2.7 Luminance, Illuminance, and Brightness 1.3 Color Image Analysis in Practical Use 1.3.1 Color Image Processing in Medical Applications 1.3.2 Color Image Processing in Food Science and Agriculture 1.3.3 Color Image Processing in Industrial Manufacturing and Nondestructive Materials Testing 1.3.4 Additional Applications of Color Image Processing 1.3.5 Digital Video and Image Databases 1.4 Further Reading 1.5 References Xlll XV 10 11 13 13 14 15 16 17 17 18 18 19 Eye and Color 2.1 Physiology of Color Vision 2.2 Receptoral Color Information 2.3 Postreceptoral Color Information 2.3.1 Neurophysiology of Retinal Ganglia Cells Reaction of Retinal Ganglia Cells to Colored Light 2.3.2 Stimuli 2.4 Cortical Color Information 2.5 Color Constant Perception and Retinex Theory 2.6 References 30 32 32 34 Color Spaces and Color Distances 3.1 Standard Color System 3.1.1 CIE Color Matching Functions 37 37 39 23 23 25 29 30 Table of Contents viii 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.1.2 Standard Color Values 3.1.3 Chromaticity Diagrams 3.1.4 MacAdam Ellipses Physics and Technics-Based Color Spaces RGB Color Spaces 3.2.1 3.2.2 CM(K) Color Space 3.2.3 YZQ Color Space 3.2.4 YUVColor Space 3.2.5 YCBCR Color Space 3.2.6 Kodak PhotoCD YClC2Color Space 3.2.7 Z,Z213 Color Space Uniform Color Spaces 3.3.1 CIELAB Color Space 3.3.2 CIELUV Color Space Perception-Based Color Spaces 3.4.1 HSZ Color Space 3.4.2 HSVColor Space 3.4.3 Opponent Color Spaces Color Difference Formulas 3.5.1 Color Difference Formulas in the RGB Color Space 3.5.2 Color Difference Formulas in the HSI Color Space 3.5.3 Color Difference Formulas in the CIELAB and CIELUV Color Spaces Color Ordering Systems 3.6.1 Munsell Color System 3.6.2 Macbeth ColorChecker 3.6.3 DIN Color Map Further Reading References Color Image Formation Technical Design of Electronic Color Cameras 4.1 4.1.1 Image Sensors 4.1.2 Mulfispectral Imaging Using Black-and-white Cameras with Color Filters 4.1.3 One-Chip CCD Color Camera 4.1.4 Three-Chip CCD Color Cameras 4.1.5 Digital Cameras Standard Color Filters and Standard Illuminants 4.2 4.2.1 Standard Color Filters 4.2.2 Standard Illuminants Photometric Sensor Model 4.3 4.3.1 Attenuation, Clipping, and Blooming 4.3.2 Chromatic Aberration Correction of the Chromatic Aberration 4.3.3 Photometric and Colorimetric Calibration 4.4 4.4.1 Nonlinearities of Camera Signals 40 41 43 44 45 48 49 50 51 52 53 53 53 55 57 58 60 62 62 63 63 64 65 66 66 67 68 69 71 71 72 74 74 76 77 78 78 80 82 83 85 87 88 89 Table of Contents 4.5 4.6 4.4.2 Measurement of Camera Linearity White Balance and Black-Level Determination 4.4.3 Transformation into the Standard Color System XYZ 4.4.4 Further Reading References Color Image Enhancement False Colors and Pseudocolors 5.I Enhancement of Real Color Images 5.2 Noise Removal in Color Images 5.3 5.3.1 Box-Filter 5.3.2 Median Filter 5.3.3 Morphological Filter Filtering in the Frequency Domain 5.3.4 5.4 Contrast Enhancement in Color Images 5.4.1 Treatment of Color Saturation and Lightness 5.4.2 Changing the Hue 5.5 References ix 01 91 93 96 06 99 100 102 102 103 104 115 I6 117 118 121 122 Edge Detection in Color Images 6.1 Vector-Valued Techniques Color Variants of the Canny Operator 6.1.1 6.1.2 Cumani Operator Operators Based on Vector Order Statistics 6.1.3 Results of Color Edge Operators 6.2 6.3 Classification of Edges 6.3.1 Physics-Based Classification 6.3.2 Classification Applying Photometric Invariant Gradients 6.4 Color Harris Operator 6.5 References 125 126 126 128 132 135 139 140 Color Image Segmentation Pixel-Based Segmentation 7.1 7.1.1 Histogram Techniques Cluster Analysis in the Color Space 7.1.2 Area-Based Segmentation 7.2 7.2.1 Region-Growing Techniques 7.2.2 Split-and-Merge Techniques Edge-Based Segmentation 7.3 7.3.1 Local Techniques 7.3.2 Segmentation by Watershed Transformation 7.3.3 Use of Watershed Transformation in Graphs 7.3.4 Expansion of the Watershed Transformation for Color Images Physics-Based Segmentation 7.4 7.4.1 Dichromatic Reflection Model 7.4.2 Classification Techniques 149 150 142 143 145 151 153 153 154 154 156 156 157 161 163 164 167 169 Table of Contents X 7.5 7.6 Comparison of Segmentation Processes References 172 i74 Highlights, Interreflections, and Color Constancy 8.1 Highlight Analysis in Color Images 8.1.1 Klinker-Shafer-Kanade Technique 8.1.2 Tong-Funt Technique 8.1.3 Gershon-Jepson-Tsotsos Technique 8.1.4 Schliins-Teschner Technique 8.1.5 Spectral Differencing Using Several Images 8.1.6 Photometric Multi-Image Technique 8.1.7 Polarization Technique 8.2 Interreflection Analysis in Color Images 8.2.1 One-Bounce Model for Interreflections 8.2.2 Determination of the One-Bounce Color Portion 8.2.3 Quarter-Circle Analysis 8.2.4 Minimization of Interreflections in Real Color Images Segmentation with Consideration to Interreflections 8.2.5 and Shadows 8.2.6 Determination of Interreflection Areas 8.2.7 Analysis of Shadow 8.2.8 Minimization of Interreflections 8.3 Color Constancy 8.3.1 Mathematical Formulation of the Color Constancy Problem 8.3.2 Techniques for Color Constancy 8.4 References 205 207 212 Static Stereo Analysis in Color Images 9.1 Geometry of a Stereo Image Acquisition System 9.2 Area-Based Correspondence Analysis Dense Disparity Maps by Block Matching 9.2.1 9.2.2 Chromatic Block Matching for Color Stereo Analysis 9.2.3 Hierarchical Block Matching in a Color Image Pyramid 9.2.4 Stereo Analysis with Color Pattern Projection 9.3 Feature-Based Correspondence Analysis 9.3.1 Edge-Based Correspondence Analysis 9.3.2 General Ideas 9.4 References 219 220 224 224 227 233 237 234 245 250 25 10 Dynamic and Photometric Stereo Analyses in Color Images 10.1 Optical Flow 10.1.1 Solution Strategy 10.1.2 Horn-Schunck Constraint for Color Image Sequences 10.2 Photometric Stereo Analysis 10.2.1 Photometric Stereo Analysis for Nonstatic Scenes 10.2.2 Photometric Stereo Analysis for Non-Lambertian Surfaces 253 254 254 255 260 26 177 177 178 181 181 183 186 188 189 194 195 197 198 200 200 201 202 202 204 263 Table of Contents 10.3 References xi 264 11 Color-Based Tracking with PTZ Cameras 267 1.1 The Background Problem 268 1.2 Methods for Tracking 270 272 11$2.1 Active Shape Models 11.2.2 Automatic Target Acquisition and Handover from Fixed to PTZ Camera 272 1.2.3 Color and Predicted Direction and Speed of Motion 273 1.3 Technical Aspects of Tracking 274 11.3.1 Feature Extraction for Zooming and Tracking 274 11.3.2 Color Extraction from a Moving Target 277 1.4 Color Active Shape Models 282 11.4.1 Landmark Points 283 1.4.2 Principal Component Analysis 283 1.4.3 Model Fitting 285 11.4.4 Modeling a Local Structure 286 11.4.5 Hierarchical Approach for Multiresolution ASM 287 1.4.6 Extending ASMs to Color Image Sequences 288 11.4.7 Partial Occlusions 294 11.4.8 Summary 296 1 S References 207 12 Multispectral Imaging for Biometrics 12.1 What is a Multispectral Image'? 12.2 Multispectral Image Acquisition 12.3 Fusion of Visible and Infrared Images for Face Recognition 12.3.1 Registration of Visible and Thermal Face Images 12.3.2 Empirical Mode Decomposition 12.3.3 Image Fusion Using EMD 12.3.4 Experimental Results 12.4 Multispectral Image Fusion in the Visible Spectrum for Face Recognition 12.4.1 Physics-Based Weighted Fusion 12.4.2 Illumination Adjustment via Data Fusion 12.4.3 Wavelet Fusion 12.4.4 CMC Measure 12.4.5 Multispectral, Multimodal, and Multi-illuminant IRIS-M3 Database 12.4.6 Experimental Results 12.5 References 301 301 302 307 309 31 313 315 Pseudocoloring in Single-Energy X-Ray Images 13.1 Problem Statement 13.2 Aspects of the Human Perception of Color 13.2.1 Physiological Processing of Color 13.2.2 Psychological Processing of Color 339 339 34 34 342 13 318 318 322 323 324 325 329 334 Table of Contents xii 13.3 13.4 13.5 13.6 13.7 13.8 13.9 Index 13.2.3 General Recommendations for Optimum Color Assignment 13.2.4 Physiologically Based Guidelines 13.2.5 Psychologically Based Guidelines Theoretical Aspects of Pseudocoloring RGB-Based Colormaps 13.4.1 Perceptually Based Colormaps 13.4.2 Mathematical Formulations HSI-Based Colormaps 13.5.1 Mapping of Raw Grayscale Data 3.5.2 Color Applied to Preprocessed Grayscale Data Experimental Results 13.6.1 Color-Coded Images Generated by RGB-Based Transforms 13.6.2 Color-Coded Images Generated by HSZ-Based Transforms Performance Evaluation 13.7.1 Preliminary Online Survey 13.7.2 Formal Airport Evaluation Conclusion References 343 344 344 345 348 348 35 354 355 357 358 359 362 367, 365 366 370 371 373 Experimental Results 361 Figure 13.18 (a) Original image and (b) histogram equalized +jet colormap Figure 13.19 Colored version of Fig 13.13 generated by the algebraic tramform of Eq 113.6) [Abi06] Figure 13.20 Colored version ofFig 13.13 generated tising the sine transform [.4bi06] 362 13 Pseudocoloring in Single-Energy X-Ray Images Figure 13.21 (a) Original and (b) sine-based color-coded image 13.6.2 Color-Coded Images Generated by HSI-Based Transforms Using the nonlinear “Springtime” color scale described in Section 13.5.1, we obtain the colored version of Fig 13.13 as shown in Fig 13.22 The colored versions of Fig 13.13 shown in Figs 13.23 and 13.24 were produced using sets “CSl” and “CS2.” A multilevel thresholding (maximum entropy approach using ICA4) was performed to segment various objects in the scene into five classes The red color was assigned to the class containing threat objects The results of various selected color assignments are shown in Figs 13.25 and 13.26 Figures 13.27 and 13.28 illustrate the results obtained from applying sets “NCS1” and “NCS2”, respectively 13.7 PERFORMANCE EVALUATION A preliminary evaluation followed by a comparative study of the various pseudocoloring methods designed and implemented were conducted in two steps The preliminary in-lab and then online surveys were carried out to refine some general aspects as to the expected outcome and adjustment of initial results A comprehensive airport testing was then conducted using actual airport screeners The results of both studies are described in the following subsections Performance Evaluation 363 Figure 13.22 Colored verJion of Fig 13.13 generated by applying the “Springtinze”color scule [Abi06] Figure 13.23 Colored version of Fig 13.13 generated b,v using set “CSI constant saturation [Abi06] ” with the Figure 13.24 Colored vmyion of Fig 13.13 generated bv using set “CS2” M,itli constani suturation (ilbiO6J 364 13 Pseudocoloring in Single-Energy X-Ray Images Figure 13.25 Color-coded version o Fig 13.13 obtained after segmentation, uJing a f constant saturation andjive different hues (complementaty colors): (a) and (b) blue, green, red, cyan, and yellow, (c) blue, yellow: cyan, red, and green [Abi06] Figure 13.26 Color-coded version of Fig 13.13 obtained after segmentation, using a constant saturation and jive different hues chosen along the color circle in clochise direction from blue to yellow: (a) and (b) blue, magenta, red, orange, and yellow, (c) blue, green, magenta, red, and yellow [Abi06] Figure 13.27 Colored version of Fig 13.13 generated by using set saturation [Abi06] VSI 'I with variable Performance Evaluation 365 Figure 13.28 Colored version of Fig 13.13 generated by using set “VS2’’ wiitli variuble satuvution [Abi06] 13.7.1 Preliminary Online Survey A preliminary online internet-based survey was conducted to compare people’s responses to grayscale-enhanced images and their color-coded counterparts Three color-coding methods and three grayscale methods were chosen for this evaluation Ten test images with low-density threat objects were selected, and 132 people responded to the survey The cosine, the HSZ histogram-based, and the rainbow maps were used for color and the intensity-stretched grayscale, negative, and histogram-equalized images for grayscale enhancements Each image was followed by three questions and at the end the overall preference among the six methods was also noted A screen shot of the survey is shown in Fig 13.29 Factors considered in this study were: The ability to detect the threat The visual appeal (how pleasantihelpful the method is) 3, The time used to identify the threat The overall preference among the given methods Each question is rated on a - 10 scale with 10 being the highest rating For the overall preference the choices were among all six different display schemes The responses were noted and plotted as shown in Fig 13.30 for item (1) and Fig 13.31 for item (4) The results showed that color-coding was significantly more effective than grayscale images in allowing people to detect threat objects in x-ray scans Eighty-six percent (86%) of the total 132 responses rated color as their preference Among the different coloring schemes, the HSV scheme developed based on the results of the study on human perception of color was ranked highest by the greatest number of people 366 13 Pseudocoloring in Single-Energy X-Ray Images Figure 13.29 Screen shot from the online survey showing the images followed questions bj) the However, the other colormaps were ranked very close to the HSV map The cosine colormap results were impressive This cosine colormap produced very continuous and smooth results when compared with the other maps Once it was established, through multiple evaluation trials with students and the staff population of the IRIS lab and with the online survey of a random population, that color-coded data is not only more appealing and boredom-proof but also more effective in detecting low-density threat objects in luggage scenes, a more formal performance study on actual airport screeners was designed and conducted 13.7.2 Formal Airport Evaluation Given the fact that airport screeners are the end users of any selected luggage coloring scheme, a natural step in this process is that the validation of the various color-coding approaches includes the responses of a representative section of the screeners’ population A fully automatic, portable, and interactive computer test was designed A snapshot of one screen of this application is shown in Fig 13.32 A set of 45 x-ray scans containing various low-density threat items in different configurations and levels of clutter were selected Performance Evaluation 367 Figure 13.30 Results of preliminan, online survey on threat detectability change with the use o color in x-ray luggage scenes f Figure 13.31 Color vs gray-level preferences by surv9) responders 368 13 Pseudocoloring in Single-Energy X-Ray Images Eight pseudocoloring approaches as described in Sections 13.4 and 13.5 were chosen according to preliminary evaluation of various transforms The selected approaches were separately applied to the luggage scans containing lowdensity threats All images were shown to screeners in random order using a random number generator, with the originals shown first in random order also This ensures that no systematic influence is gained by colored images over the noncolored images in the detection of threats The screeners were asked not only to affirm seeing a threat but also to point and click on the threat to ensure they saw the actual threat The screeners were also asked to rank the images in terms of their visual clarity and ease of interpretation, which is an important fact in relieving boredom and keeping the screener’s level of concentration relatively high The evaluation sessions were conducted at McGhee Tyson Airport in Knoxville, Tennessee and involved a total of 40 Transportation Security Administration (TSA) luggage screeners Five types of information were collected for each image shown: Did the screener see a threat object in the image? If so, how many (1 or 2) threat items were seen? Could the screener correctly click on the location of at threat object? If two threat items were indicated, could the screener correctly click on the location of the second threat object? A rating (from to 10, with 10 being best) of how helpful the screener believed the displayed image was in detecting the threat object Figure 13.32 Screen shot of’ the graphical user interface rmd in the airport e\dirtitiori s t z ~ & (A bi061 Performance Evaluation 369 After all enhanced images had been shown as single windows, a montage image was shown for each original In this montage the original image is shown side by side (for comparison purposes) with each of its colored versions and the screener is asked to rate each of the nine images on a 1-to-10 scale with 10 being the best in terms of ease of threat detection in the image An example of one montage window is shown in Fig 13.33 The ability of the screener to correctly click on the threat item location within the luggage was determined through use of a binary mask image When the screener clicks on a specific ( x , y ) location in the image being evaluated, the program checks the same location in the corresponding mask image If this pixel location has a value of the answer is recorded as being correct Once all screeners completed the evaluation, the composite set of data obtained was evaluated in Excel to determine what trends might be apparent Figure 13.34 presents a graphic showing the percent of all screeners who were able to correctly click on the threat location for the original image and for each of the color enhancement methods This graphic indicates that a low percentage (3 %) of screeners were able to correctly locate the threat object in the original grayscale luggage scan On the other hand, the color-enhanced images faired much better, ranging from 56.5 to 69.5% recognition The image with enhancement method “Heqstrmap4” had the highest recognition rate This enhancement consists of histogram equalization, followed by image stretching, followed by application of an in-house-developed colormap called “map4.” Figure 13.33 Montage of original and all color-coded images for comparative rating [Abi06] 70 13 Pseudocoloring in Single-Energy X-Ray Images Figure 13.34 Percentage of screeners able to correctly identifi threat objects on each type of image The other important evaluation criterion collected was the rating of each type of image color coding Again, the original image scored lowest (1.64) while the enhanced images were all significantly higher, being in the range of 2.56 to 5.24 Color scale “Springtime” again provided the highest rating Overall, the results obtained from the screener evaluations indicated a clear preference for colorenhanced x-ray scans over original, raw grayscale luggage scans as supplied from the scanning equipment Four approaches, “Warm,” “Springtime,” “CS 1,” and “CS2,” received higher average ratings than others Of the four approaches, all except “Warm” were designed based on the HSI color model, which confirms earlier remarks that the HSZ color model is more suitable to human interpretation and therefore more effective in revealing low-density threats concealed in x-ray luggage scans in this case 13.8 CONCLUSION A number of novel color transforms were introduced, applied to luggage scenes, and tested by screeners in an airport environment Proper colormapping schemes have been designed based on perceptive and cognitive features without which it is impossible to produce an effective visualization The expectation that pseudocoloring techniques can provide additional enhancement of x-ray luggage scans, better data visualization, increased screeners alertness, and longer References 371 attentioniretention was demonstrated by experimental results and evaluations by actual airport screeners It was shown through visual interpretation, and more importantly through testing on TSA airport screeners, that newly developed colormapping techniques are very valuable tools in increasing the rate o f low-density threat detection in xray luggage scans A significant increase of up to 97%, as compared with results from the original data, in the rate of threat detection was obtained when colorcoded data was used by screeners Feedback from screeners also rated the colorprocessed data, on the average, as 219% more helpful in detecting a threat than the raw data Not only did the testing show that color-processed data is more effective than grayscale data in detecting threats and keeping the screener’s attention, but we were able to also rank the set of colormapping procedures as to which is most effective and most appealing to screeners In comparing the RGB-based approaches with the HSZ-based approaches, this latter color space proved superior, which was expected given the many known advantages of the HSI space in human-based applications [Wei97] Experimental results show that the HSZ-based methods produce results consistent with the human assessment Future efforts would include more testing with the introduction of images containing no threat into the set o f images evaluated to study performance in terms of the rate of false positives 13.9 [Abi041 [Abi06] [But021 [Cla89] [Cze99] [Dai96] [Gon02] [Hea96] REFERENCES B Abidi, M Mitckes, J Liang, M Abidi Improving the detection of low density weapons in x-ray luggage scans using image enhancement and novel scene decluttering techniques J Electronic Imaging 13 (2004), pp 523-538 B Abidi, Y Zheng, A.V Gribok, M.A Abidi Improving weapon detection in single energy x-ray images through pseudocoloring IEEE Transactions on Systems, Man, 13Cybernetics, Part C: Aplications and Reviews 36 (2006), pp 184-796 V Butler, R.W Poole, Jr Rethinking checked-baggage screening Policy Study No 297, Reason Public Policy Institute, July 2002 F.J Clarke, J.K Leonard Proposal for a standardized continuous pseudocolor spectrum with optimal visual contrast and resolution Proc 3rd Int Conference on Image Processing and its Application, pp 687-691, 1989 R.N Czenvinski, D.L Jones, W.D O’Brien, Jr Detection of lines and boundaries in speckle images - Application to medical ultrasound IEEE Transactions on Medical Imaging 18, (l999), pp 126-136 J Dai, S Zhou Computer Aided Pseudocolor coding of gray image: Complementary color coding technique Proc SPIE 2898, pp 186-191, 1996 R.C Gonzalez, R.E Woods Digital Image Processing 2nd ed., PrenticeHall, Upper Saddle River, New Jersey, 2002 C.G Healey Choosing effective colours for data visualization Proc IEEE Visualization, pp 263-270, 1996 72 [Lev921 [Mac991 [May 85 [Mur84] [Oue04] [War881 [Wei97] [Wri97] 13 Pseudocoloring in Single-Energy X-Ray Images H Levkowitz, G Herman Color scales for image data I€E€ Computer Graphics and Applications 12 (1992), pp 72-80 L.W MacDonald Using color effectively in computer graphics IEEE Computer Graphics andApplications 19 (1999), pp 20-35 W.T Mayo, P.V Shankar, L.A Ferrari Color-coding medical ultrasonic images with frequency information Proc SPI€ 575, pp 255-261, 1985 G.M Murch Physiological principles for the effective use of volor IEEE Computer Graphics and Applications (1 984), pp 49-54 N Ouerhani, R Wartburg, H Hugli, R Muri Empirical validation of the saliency-based model of visual attention, Electronics Letters on Computer Vision and Image Analysis (2004), pp 13-24 X.Q Shi, P Sallstrom, U Welander A color coding method for radiographic images Image and Vision Computing 20 (2002), pp 761-767 J.M Taylor, G.M Murch The effective use of color in visual displays: Text and graphics applications Color Research and Applications 11 ( 986) Supplement pp S3-10 C Ware Color sequences for univariate maps: Theory, experiments and principles I€€€ Computer Graphics and Applications (1988), pp 41-49 G.Q Wei, K Arbter, G Hirzinger Automatic tracking of laproscopic instruments by color coding Lecture Notes in Computer Science 1205 (1997) P Wright, D Mosser-Wooley, B Wooley Techniques & tools for using color in computer interface design ACM CrossRoads (Spring 1997) Digital Color Image Processing by Andreas Koschan and Mongi Abidi Copyright 2008 John Wiley & Sons, Inc INDEX Achromates, 87 Acousto-optic tunable filter, 303 Active camera, 254 Albedo, 13 Aperture problem, 254 Area-based stereo, 224 ASM, 282 Black level, 92 Block matching, 225 Blooming, 85 Body color, 80 Box-Filter, 103 Brightness, 14, 25 photometric, 13 Canny operator, 126, 138 Chromatic aberration axial, 85 lateral, 85 Chromaticity, Chromaticity diagram, 41 Chromatopsy, 23 CIE, 37 chromaticity diagram, 42 standard observer, 39 CIELAB, 53 CIELUV, 55 CMy(K), 48 Color stimulus function, 40 Color blindness, 28 Color constancy, 10, 32,204 retinex theory, 207 supervised, 209 Color edge, Color gamut, 42 Color histogram, 169 Color image segmentation, 149 Color management, 46 Color mixture additive, subtractive, 6,27 Color space, 37 CIELAB, 53 CIELUV, 5 CMY(K), HSI, 58 HSV, 60 z l I I , 53 RGB, 45 sRGB, 47 XYZ, 41 YC,C*,52 YCBCR,51 YZQ, 49 YUV, 50 Colormaps HSI-based, 354 perceptually based, 348 RGB-based, 348 Contrast, 11 relative brightness contrast, 1 relative saturation contrast, 1 simultaneous brightness contrast, 11 successive color contrast, 12 Contrast enhancement, 18 373 374 Correspondence analysis area-based, 224 feature-based, 244 Cumani operator, 128, 138 , 81 Dense disparity map, 224 Derivative of a color image, Dichromates, 27 Dichromatic plane, 168 Dichromatic reflection model, 167 Disparity, 222 Distance Euclidean, 63 geodesic, 158 DRM, 167 Edge detection, 125 Empirical mode decomposition, 1 Epipolar line, 222 False-color image, Feature-based stereo, 244 Filter Kodak Wratten, 78 Four-color theory, 29 Fresnel reflection, 167 Functional matrix Gamma, 89 Geodesic distance, 158 Geometrical image modification, 99 Gradient, Gradient vector, Harris operator, 143 color, 144 Horn-Schunck constraint, 256 color images, 257 HSI, 58 HSV, 60 Hue, 25 Hyperspectral image, 302 Illuminance, 13 Illuminant standard, 80 Index Illumination adjustment, 322 Image band, 301 channel, 301 false-color, hyperspectral, 302 hyperspectral, multichannel, multispectral, 8, 302 pixel, pseudocolor, quantization, resolution, size, Image restoration, 99 Image retrieval, 18 Indexed color, Interest point detector, 143 Interface reflection, 167 Interferometer type filters, 303 Interreflection, 194 analysis, 195 Intrinsic mode functions, 1 Iterative closest point, 10 Jacobian matrix, JND, 347 Landmark points, 283 Lightness, 14 Liquid crystal tunable filter, 304 LOG-Filter, 245 Lookup table, 91 Luminance, 13,25 MacAdam ellipses, 44 Macbeth ColorChecker, 1,94,209 Macbeth-ColorChecker, 94 Macropixel, 75 Metameric, 27 Mexican Hat operator, 136 Minimum vector dispersion edge detector, 135 Mondrian image, 33 Mondrian images, 1 Monochromates, 28 Mosaic filter, 74 Index Motion vector, 224 Multispectral image, 301 Mutual illumination 194 NIRM, 168 OFF-center neurons, 30 ON-center neurons, 30 One-bounce model, 196 Opponent color space, 62 Opponent color theory, 29 Optical density, 91 Optical flow, 254 PCA algorithm, 284 Photometric compatibility constraint, 224 Photometric stereo, 261 Pixel, Principal components analysis, 308 Pseudocolor image, PTZ camera, 267 Purple boundary, 42 Receptive field, 30 Reflectance, 13 Region, 149 Retinex theory, 33, 207 RGB, 45 Saturation, 25 Sensor frame transfer, 73 interline transfer, 72 375 Shadow analysis, 202 Smoothness constraint, 95 Spectral color transmission, Spectral differencing, 186 Spectroradiometer, 305 sRGB, 47 Standard color values, 40 Standard illuminant, 80 A , 81 C, 81 , 81 Standard stereo geometry, 223 Stereo feature-based, 244 photometric, 261 Stereo vision, 219 Surface reflection, 167 Trichromates, 27 Trichromatic theory, 26 True-color image, Vector dispersion edge detector, 134 Vision mesopic, 26 scotopic, 26 Wavelength complementary, 44 White balance, 92 YCIC,, 52 YCBC,, 51 YIQ, 49 YUV, 50 ... Cataloging-in-Publication Data: Koschan, Andreas, 195 6Digital color image processing / b y Andreas Koschan and Mongi A Abidi p, cm ISBN 97 8-0 -4 7 0- 1470 8-5 (cloth) Image processing -Digital techniques Color. .. in Color Image Processing 1.2.1 What Is a Digital Color Image? 1.2.2 Derivative of a Color Image 1.2.3 Color Edges 1.2.4 Color Constancy 1.2.5 Contrast of a Color Image 1.2.6 Noise in Color Images... in color image processing is established 1 Introduction 1.2.1 What Is a Digital Color Image? The central terminology of color image processing is that of the digital color image A digital image