SIST ISO 20462-3:2011

2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.VOLNRYQHPhotographie - Méthodes psychophysiques expérimentales pour estimer la qualité d'image - Partie 3: Méthode <>Photography - Psychophysical experimental methods for estimating image quality - Part 3: Quality ruler method37.040.01Fotografija na splošnoPhotography in generalICS:Ta slovenski standard je istoveten z:ISO 20462-3:2005SIST ISO 20462-3:2011en01-julij-2011SIST ISO 20462-3:2011SLOVENSKI
STANDARD



SIST ISO 20462-3:2011



Reference numberISO 20462-3:2005(E)© ISO 2005
INTERNATIONAL STANDARD ISO20462-3First edition2005-11-01Photography — Psychophysical experimental methods for estimating image quality — Part 3: Quality ruler method Photographie — Méthodes psychophysiques expérimentales pour estimer la qualité d'image — Partie 3: Méthode «quality ruler»
SIST ISO 20462-3:2011



ISO 20462-3:2005(E) PDF disclaimer This PDF file may contain embedded typefaces. In accordance with Adobe's licensing policy, this file may be printed or viewed but shall not be edited unless the typefaces which are embedded are licensed to and installed on the computer performing the editing. In downloading this file, parties accept therein the responsibility of not infringing Adobe's licensing policy. The ISO Central Secretariat accepts no liability in this area. Adobe is a trademark of Adobe Systems Incorporated. Details of the software products used to create this PDF file can be found in the General Info relative to the file; the PDF-creation parameters were optimized for printing. Every care has been taken to ensure that the file is suitable for use by ISO member bodies. In the unlikely event that a problem relating to it is found, please inform the Central Secretariat at the address given below.
©
ISO 2005 All rights reserved. Unless otherwise specified, no part of this publication may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying and microfilm, without permission in writing from either ISO at the address below or ISO's member body in the country of the requester. ISO copyright office Case postale 56 • CH-1211 Geneva 20 Tel. + 41 22 749 01 11 Fax + 41 22 749 09 47 E-mail copyright@iso.org Web www.iso.org Published in Switzerland
ii
© ISO 2005 – All rights reserved
SIST ISO 20462-3:2011



ISO 20462-3:2005(E) © ISO 2005 – All rights reserved
iiiContents Page Foreword.iv Introduction.v 1 Scope.1 2 Normative references.1 3 Terms and definitions.1 4 Quality ruler experiments.5 4.1 General properties of quality rulers.5 4.2 Experimental conditions and reported results.5 4.3 Attributes varied in quality rulers.5 5 Hardcopy quality ruler implementation.6 5.1 Physical apparatus.6 5.2 Reference stimuli.7 6 Softcopy quality ruler implementation.8 6.1 Physical apparatus.8 6.2 Reference stimuli.8 6.3 Controlling software.8 7 Generation of quality ruler stimuli.9 7.1 General requirements.9 7.2 Modulation transfer functions (MTFs).9 7.3 Scene-dependent ruler calibration.11 8 Standard quality scale (SQS) determinations.12 8.1 Properties of the SQS.12 8.2 Experimental requirements.12 Annex A (informative) Sample instructions for a hardcopy quality ruler experiment.13 Annex B (informative) Sample instructions for a softcopy quality ruler experiment.15 Annex C (informative) Sample code of a binary search routine for the softcopy quality ruler.17 Annex D (informative) Calibration of the standard quality scale (SQS) and its reference stimuli.18 Annex E (informative) Example of results from quality ruler experiments.20 Bibliography.24
SIST ISO 20462-3:2011



ISO 20462-3:2005(E) iv
© ISO 2005 – All rights reserved Foreword ISO (the International Organization for Standardization) is a worldwide federation of national standards bodies (ISO member bodies). The work of preparing International Standards is normally carried out through ISO technical committees. Each member body interested in a subject for which a technical committee has been established has the right to be represented on that committee. International organizations, governmental and non-governmental, in liaison with ISO, also take part in the work. ISO collaborates closely with the International Electrotechnical Commission (IEC) on all matters of electrotechnical standardization. International Standards are drafted in accordance with the rules given in the ISO/IEC Directives, Part 2. The main task of technical committees is to prepare International Standards. Draft International Standards adopted by the technical committees are circulated to the member bodies for voting. Publication as an International Standard requires approval by at least 75 % of the member bodies casting a vote. ISO 20462-3 was prepared by Technical Committee ISO/TC 42, Photography. ISO 20462 consists of the following parts, under the general title Photography — Psychophysical experimental methods for estimating image quality: ⎯ Part 1: Overview of psychophysical elements ⎯ Part 2: Triplet comparison method ⎯ Part 3: Quality ruler method
SIST ISO 20462-3:2011



ISO 20462-3:2005(E) © ISO 2005 – All rights reserved
vIntroduction There are many circumstances under which it is desirable to quantify image quality in a standardized fashion that facilitates interpretation of results within a given experiment and/or comparison of results between different experiments. Such information can be of value in assessing the performance of different capture or display devices, image processing algorithms, etc. under various conditions. However, the choice of the best psychometric method for a particular application may be difficult to make, and interpretation of the rating scales produced by the numerical analyses is frequently ambiguous. Furthermore, none of the commonly used rating techniques provides an efficient mechanism for calibration of the results against a standardised numerical scale or associated physical references, which is desirable when results of different experiments are to be compared or integrated. The three parts of ISO 20462 address the need for documented means of determining image quality in a calibrated fashion. Part 1 provides an overview of practical psychophysics and aids in identifying the better choice between the two alternative approaches described in Part 2 (triplet comparison method[1][2]) and Part 3 (quality ruler method[3]). These two techniques are complementary and together are sufficient to span a wide range of practical applications. Parts 2 and 3 document both specific experimental methods and associated data reduction techniques. It is the intent of these methods to produce results that are not merely directional in nature, but are expressed in terms of relative or fixed scales that are calibrated in terms of just noticeable differences (JNDs), so that the significance of experimentally measured stimulus differences is readily ascertained. The quality ruler method described in this part of ISO 20462 is particularly suitable for measuring quality differences exceeding one JND. The ratings given by an observer can be converted to JND values in real time, rather than having to wait until the entire experimental data set has been collected and analysed. Furthermore, with suitable reference stimuli, the quality ruler method permits the results to be reported using the standard quality scale (SQS), a fixed numerical scale that: a) is anchored against physical standards; b) has one unit corresponding to one JND; and c) has a zero point corresponding to an image having little identifiable information content. Reflection prints calibrated against the absolute SQS, which are referred to as standard reference stimuli (SRS), will be available on the I3A website. This part of ISO 20462 also describes how users can conveniently generate their own quality ruler images with correct relative calibrations and, if desired, calibrate them absolutely against the SRS. The International Organization for Standardization (ISO) draws attention to the fact that it is claimed that compliance with this document may involve the use of US Patent Numbers 6,639,999 and 6,658,139 concerning the quality ruler given in Clauses 4 to 6. ISO takes no position concerning the evidence, validity and scope of this patent right. The holder of this patent right has assured ISO that he is willing to negotiate licences under reasonable and non-discriminatory terms and conditions with applicants throughout the world. In this respect, the statement of the holder of this patent right is registered with ISO. Patent inquiries may be addressed to: General Council and Senior Vice President Eastman Kodak Company 345 State Street Rochester, NY 14650 USA SIST ISO 20462-3:2011



ISO 20462-3:2005(E) vi
© ISO 2005 – All rights reserved Attention is drawn to the possibility that some of the elements of this document may be the subject of patent rights other than those identified above. ISO shall not be held responsible for identifying any or all such patent rights. SIST ISO 20462-3:2011



INTERNATIONAL STANDARD ISO 20462-3:2005(E) © ISO 2005 – All rights reserved
1Photography — Psychophysical experimental methods for estimating image quality — Part 3: Quality ruler method 1 Scope This part of ISO 20462 specifies: a) the nature of a quality ruler; b) hardcopy and softcopy implementations of quality rulers; c) how quality rulers may be generated or obtained; and d) the standard quality scale (SQS), a fixed numerical scale that may be measured using quality rulers. 2 Normative references The following referenced documents are indispensable for the application of this document. For dated references, only the edition cited applies. For undated references, the latest edition of the referenced document (including any amendments) applies. ISO 3664, Viewing conditions — Graphic technology and photography 3 Terms and definitions For the purposes of this document, the following terms and definitions apply. 3.1 artefactual attribute attribute of image quality that, when evident in an image, nearly always leads to a loss of overall image quality EXAMPLE Examples of artefactual attributes include noise and aliasing. NOTE The commonly used terms defect and impairment are similar in meaning. 3.2 attribute aspect, dimension, or component of overall image quality cf. artefactual attribute (3.1) and preferential attribute (3.10) EXAMPLE Examples of image quality attributes include image structure properties such as sharpness and noise; colour and tone reproduction properties such as contrast, colour balance, and relative colourfulness; and digital artefacts such as aliasing, contouring, and compression defects. SIST ISO 20462-3:2011



ISO 20462-3:2005(E) 2
© ISO 2005 – All rights reserved 3.3 image quality impression of the overall merit or excellence of an image, as perceived by an observer neither associated with the act of photography nor closely involved with the subject matter depicted NOTE The purpose of defining image quality in terms of third-party (uninvolved) observers is to eliminate sources of variability that arise from more idiosyncratic aspects of image perception and pertain to attributes outside the control of imaging system designers. 3.4 instructions set of directions given to the observer for performing the psychophysical evaluation task 3.5 just noticeable difference JND stimulus difference that leads to a 75:25 proportion of responses in a paired comparison task cf. quality JND (3.12) 3.6 magnitude estimation method psychophysical method involving the assignment of a numerical value to each test stimulus that is proportional to image quality; typically, a reference stimulus with an assigned numerical value is present to anchor the rating scale NOTE The numerical scale resulting from a magnitude estimation experiment is usually assumed to constitute a ratio scale which, ideally, is a scale in which a constant percentage change in value corresponds with one JND. In practice, modest deviations from this behaviour occur, complicating the transformation of the rating scale into units of JNDs without inclusion of unidentified reference stimuli (having known quality) among the test stimuli. 3.7 multivariate describing a series of test or reference stimuli that vary in multiple attributes of image quality 3.8 observer individual performing the subjective evaluation task in a psychophysical method 3.9 paired comparison method psychophysical method involving the choice of which of two simultaneously presented stimuli exhibits greater or lesser image quality or an attribute thereof, in accordance with a set of instructions given to the observer NOTE Two limitations of the paired comparison method are as follows. a) If all possible stimulus comparisons are done, as is usually the case, a large number of assessments are required for even modest numbers of experimental stimulus levels [if N levels are to be studied, N (N − 1)/2 paired comparisons are needed]. b) If a stimulus difference exceeds approximately 1,5 JNDs, the magnitude of the stimulus difference cannot be directly estimated reliably because the response saturates as the proportions approach unanimity. However, if a series of stimuli having no large gaps are assessed, the differences between more widely separated stimuli may be deduced indirectly by summing smaller, reliably determined (unsaturated) stimulus differences. The standard methods for transformation of paired comparison data to an interval scale (a scale linearly related to JNDs) perform statistically optimized procedures for inferring the stimulus differences, but they may yield unreliable results when saturated responses are included in the analysis. SIST ISO 20462-3:2011



ISO 20462-3:2005(E) © ISO 2005 – All rights reserved
33.10 preferential attribute attribute of image quality that is invariably evident in an image, and for which the preferred degree is a matter of opinion, depending upon both the observer and the image content EXAMPLE Examples of preferential image quality attributes include colour and tone reproduction properties such as contrast and relative colourfulness. NOTE 1 Because the perceived quality associated with a preferential attribute is dependent upon both the observer and image content, in studies involving variations of preferential attributes, particular care is needed in the selection of representative sets of stimuli and groups of observers. NOTE 2 The term noticeable in just noticeable difference is not linguistically strictly correct when applied to a preferential attribute, but is nonetheless retained in this part of ISO 20462 for convenience. For example, the higher contrast stimulus of a pair differing only in contrast might be readily identified by all observers, whereas there might be a lack of consensus regarding which of the two images was higher in overall image quality. Nonetheless, if the responses from the paired comparison for quality were in the proportion of 75:25, the image chosen more frequently would be said to be one JND higher in quality. The JND is best regarded as a measurement unit tied to the predicted or measured outcome of a paired comparison. 3.11 psychophysical method experimental technique for subjective evaluation of image quality or attributes thereof, from which stimulus differences in units of JNDs may be estimated cf. magnitude estimation (3.6), paired comparison (3.9), quality ruler (3.13), and triplet comparison methods (3.22) 3.12 quality just noticeable difference quality JND measure of the significance or importance of quality variations, corresponding to a stimulus difference that leads to a 75:25 proportion of responses in a paired comparison task in which multivariate stimuli pairs are assessed in terms of overall image quality NOTE See attribute JND (3.3) and quality JND (3.14) in ISO 20462-1:— for greater detail. 3.13 quality ruler method psychophysical method that involves quality or attribute assessment of a test stimulus against a series of ordered, univariate reference stimuli that differ by known numbers of JNDs 3.14 reference stimulus image provided to the observer for the purpose of anchoring or calibrating the perceptual assessments of test stimuli in such a manner that the given ratings may be converted to JND units NOTE The plural is reference stimuli. 3.15 scene content or subject matter of an image, or a starting image from which multiple stimuli may be produced through different experimental treatments NOTE Typically, stimuli depicting the same scene are compared in a psychophysical experiment, because it is the effect of the treatment that is of interest, and differences in image content could cause spurious effects. In cases where scene content is not matched, a number of scenes should be used so that scene effects may be expected to average out. SIST ISO 20462-3:2011



ISO 20462-3:2005(E) 4
© ISO 2005 – All rights reserved 3.16 standard quality scale
SQS fixed numerical scale of quality having the following properties: a) the numerical scale is anchored against physical standards; b) a one unit increase in scale value corresponds to an improvement of one JND of quality; and c) a value of zero corresponds to an image having so little information content that the nature of the subject of the image is difficult to identify 3.17 standard reference stimuli SRS set of reflection prints used in the hardcopy quality ruler, which vary in sharpness and are calibrated against the standard quality scale (SQS) NOTE The SRS will be available on the I3A web site. 3.18 stimulus image presented or provided to the observer either for the purpose of anchoring a perceptual assessment (a reference stimulus) or for the purpose of subjective evaluation (a test stimulus) NOTE The plural is stimuli. 3.19 suppression perceptual effect in which one attribute is present in a degree that seriously degrades image quality and thereby reduces the impact that other attributes have on overall quality, compared to the impact they would have had in the absence of the dominant attribute NOTE To generate reference stimuli that are separated by a specified number of JNDs based on variations in one attribute, it will be necessary to ensure that other attributes do not significantly suppress the impact of the attribute varied. 3.20 test stimulus image presented to the observer for subjective evaluation NOTE The plural is test stimuli. 3.21 treatment controlled or characterized source of the variations between test stimuli (excluding scene content) that are to be investigated in a psychophysical experiment EXAMPLE Examples of treatments include different image processing algorithms, variations in capture or display device properties, changes in image capture conditions (e.g. camera exposure), etc. NOTE Different treatments may be achieved through hardware or software changes, or may be numerical simulations of such effects. Typically, a series of treatments is applied to multiple scenes, each generating a series of test stimuli. The effect of the treatment may then be determined by averaging the results over scene and observer to improve signal to noise and reduce the likelihood of systematic bias. 3.22 triplet comparison psychophysical method that involves the simultaneous scaling of three test stimuli with respect to image quality or an attribute thereof, in accordance with a set of instructions given to the observer NOTE The triplet comparison method is described in more detail in ISO 20462-2. SIST ISO 20462-3:2011



ISO 20462-3:2005(E) © ISO 2005 – All rights reserved
53.23 univariate describing a series of test or reference stimuli that vary only in a single attribute of image quality 4 Quality ruler experiments 4.1 General properties of quality rulers A quality ruler is a univariate series of reference stimuli depicting the same scene and having known stimulus differences expressed in JNDs of quality. The reference stimuli are presented to the observer in a fashion facilitating: a) the identification of the reference stimuli closest in quality to the test stimulus; and b) the comparison of the test stimulus to those reference stimuli under rigorously matched viewing conditions. Both hardcopy (Clause 5) and softcopy (Clause 6) implementations of quality rulers are described in this standard. Ruler images may be generated by the user (Clause 7). Reflection prints varying in sharpness and calibrated against the standard quality scale (SQS) are referred to as standard reference stimuli (SRS) (Clause 8). NOTE The SRS will be available on the I3A web site. The SRS may be used as ruler images or used to calibrate user-generated ruler images on an absolute basis, as distinguished from the relative calibration described in Clause 7. 4.2 Experimental conditions and reported results Requirements regarding observer selection, test stimulus properties, instructions to the observer, viewing conditions, and reporting of results are set forth in ISO 20462-1. NOTE 1 Sample instructions to the observer for hardcopy and softcopy quality ruler experiments are provided in informative Annexes A and B, respectively. The viewing requirements of ISO 3664 shall be met except as modified in 4.4 of ISO 20462-1:—. Reported values of quality in JNDs or SQS units shall be specifically identified if they are calculated from data 20 % or more of which fall at one of the ends of, or outside, the range of the quality ruler from which they were derived. NOTE 2 Values based on ratings outside the range of the ruler will be less reliable because of extrapolation effects. In addition, when test samples fall within a JND or two of the high quality end of the ruler, a slight bias may result from observers avoiding use of ratings outside the ruler range. When preferential attributes (e.g. of colour and tone reproduction) are assessed using a quality ruler, it may be desirable to degrade all the test stimuli slightly by blurring (in the case of a ruler varying in sharpness) to allow headroom for test stimuli that are preferred over the reference stimulus. The pedigree of the rulers used shall be reported, which entails specifying whether they are standard reference stimuli (SRS) or were otherwise generated. If the latter, the attribute varied in the rulers shall be stated. If such rulers vary in sharpness, the method of calibration shall be stated, which shall either be by comparison with SRS or using the average scene relationship (see 7.2). 4.3
Attributes varied in quality rulers Clause 7 describes the generation of reference stimuli for rulers varying in sharpness, through modification of the modulation transfer function (MTF) of the system generating the images. Quality rulers may alternatively vary in other attributes, although only one attribute shall change within a given ruler. Alternative attributes that are varied in a quality ruler should be artefactual in nature. SIST ISO 20462-3:2011



ISO 20462-3:2005(E) 6
© ISO 2005 – All rights reserved NOTE The variation of preferential attributes within quality rulers is discouraged because of the additional variability associated with such attributes. Sharpness has been selected as the reference attribute because of several desirable characteristics: a) it is easily manipulated through image processing; b) it is correlated with MTF, which is readily determinable; c) it has low scene and observer variability; and d) it exerts a strong influence on quality in practical imaging systems. Quality rulers varying in attributes other than sharpness shall be calibrated by having their reference stimuli rated against quality rulers varying in sharpness and meeting the criteria stated in this part of ISO 20462. The calibration experiment shall meet the specifications set forth in ISO 20462-1 and this part ISO 20462, with the exception that data from a minimum of 20 observers shall be averaged to determine the calibration. 5 Hardcopy quality ruler implementation 5.1 Physical apparatus The hardcopy quality ruler apparatus shall consist of the following: a) a sliding or translating fixture onto or into which a series of reference stimuli may be mounted or inserted (the ruler); b) a test stimulus fixture in close proximity to the ruler; c) a base surface upon which the ruler and the test stimulus fixture are attached; d) an illumination system; and e) a headrest or other device constraining the viewing distance (the distance from the observer’s eye to the test and reference stimuli). The ruler shall be constructed so that the observer may easily slide it to bring any of two reference stimuli into direct comparison with the test stimulus. In this triangular configuration of one test stimulus and two reference stimuli, the illumination level, illumination angle, viewing distance, and viewing angle shall be sensibly matched between the three stimuli. These features are illustrated in Figure 1. The illumination angle should be 45° and shall fall between 30° and 60°. The viewing distance to any of the three stimuli shall be constrained by the headrest or equivalent mechanism to a range not exceeding 4 % of the value of the arithmetic average viewing distance. The range of the viewing distances of the three stimuli at a given observer head position shall not exceed 2 % of the arithmetic average viewing distance. The viewing angle should be normal to the stimulus surfaces and shall be within 10° of being perpendicular. Specular reflections from the stimuli shall not be visible from the observer’s position. NOTE
Achieving the closely matched viewing conditions of the test stimulus and the two reference (ruler) stimuli in the triangular configuration (which facilitates rating interpolation by the observer) is simplified if the physical separation of the three stimuli is minimized. Because some rulers may contain landscape (horizontal) format images and others portrait format (vertical) images, it may be advantageous for the test stimulus fixture to translate vertically. To match viewing angles between the test and reference stimuli, the receiving surface of the test stimulus fixture may have to be tilted. SIST ISO 20462-3:2011



ISO 20462-3:2005(E) © ISO 2005 – All rights reserved
75.2 Reference stimuli The reference stimuli shall be ordered from highest to lowest quality from left to right in a horizontally translating ruler or top to bottom in a vertically translating ruler. These stimuli should be spaced by increments of approximately three JNDs. Each stimulus shall be labelled with an integer, and the observer shall provide ratings interpolated to the nearest integer value, which should correspond to approximately one JND scale resolution. The integer labels shall be chosen so that negative ratings are unlikely. NOTE 1 The use of two interpolating positions between stimuli (for example, stimuli labelled 3 units apart with interpolation to one unit) has been found to yield a uniform and unbiased use of the numerical ratings, whereas when
...