ASTM E1808-1996(2003) Standard Guide for Designing and Conducting Visual Experiments《设计和操纵可视实验标准指南》.pdf

1、Designation: E 1808 96 (Reapproved 2003)Standard Guide forDesigning and Conducting Visual Experiments1This standard is issued under the fixed designation E 1808; the number immediately following the designation indicates the year oforiginal adoption or, in the case of revision, the year of last revi

2、sion. A number in parentheses indicates the year of last reapproval. Asuperscript epsilon (e) indicates an editorial change since the last revision or reapproval.1. Scope1.1 This guide is intended to help the user decide on the typeof viewing conditions, visual scaling methods, and analysisthat shou

3、ld be used to obtain reliable visual data.1.2 This guide is intended to illustrate the techniques thatlead to visual observations that can be correlated with objectiveinstrumental measurements of appearance attributes of objects.The establishment of both parts of such correlations is anobjective of

4、Committee E12.1.3 Among ASTM standards making use of visual observa-tions are Practices D 1535, D 1729, D 3134, D 4086, andE 1478; Test Methods D 2616, D 3928, and D 4449; and GuideE 1499.1.4 This guide does not purport to address all of the safetyconcerns, if any, associated with its use. It is the

5、 responsibilityof the user of this standard to establish appropriate safety andhealth practices and determine the applicability of regulatorylimitations prior to use.2. Referenced Documents2.1 ASTM Standards:2D 1535 Practice for Specifying Color by the Munsell Sys-temD 1729 Practice for Visual Appra

6、isal of Colors and ColorDifferences of Diffusely Illuminated Opaque MaterialsD 2616 Test Method for Evaluation of Visual Color Differ-ence with a Gray ScaleD 3134 Practice for Establishing Color and Gloss Toler-ancesD 3928 Test Method for Evaluation of Gloss or SheenUniformityD 4086 Practice for Vis

7、ual Evaluation of MetamerismD 4449 Test Method for Visual Evaluation of Gloss Differ-ences Between Surfaces of Similar AppearanceE 284 Terminology of AppearanceE 1478 Practice for Visual Color Evaluation of TransparentSheet MaterialsE 1499 Guide to the Selection, Evaluation, and Training ofObservers

8、3. Terminology3.1 The terms and definitions in Terminology E 284 areapplicable to this guide.3.2 Definitions:3.2.1 appearance, nin psychophysical studies, perceptionin which the spectral and geometric aspects of a visual stimulusare integrated with its illuminating and viewing environment.3.2.2 obse

9、rver, none who judges visually, qualitatively orquantitatively, the content of one or more appearance attributesin each member of a set of stimuli.3.2.3 sample, na small part or portion of a material orproduct intended to be representative of the whole.3.2.4 scale, vto assess the content of one or m

10、ore appear-ance attributes in the members of a set of stimuli.3.2.4.1 DiscussionAlternatively, scales may be deter-mined by assessing the difference in content of an attributewith respect to the differences in that attribute among themembers of the set.3.2.5 specimen, na piece or portion of a sample

11、 used tomake a test.3.2.6 stimulus, nany action or condition that has thepotential for evoking a response.3.3 Definitions of Terms Specific to This Standard:3.3.1 anchor, nthe stimulus from which a just-perceptibledifference is measured.3.3.2 anchor pair, na pair of stimuli differing by a definedamo

12、unt, to which the difference between two test stimuli iscompared.3.3.3 interval scale, na scale having equal intervals be-tween elements.3.3.3.1 DiscussionLogical operations such as greater-than, less-than, equal-to, and addition and subtraction can beperformed with interval-scale data.3.3.4 law of

13、comparative judgmentsan equation relatingthe proportion of times any stimulus is judged greater, accord-ing to some attribute, than any other stimulus in terms ofjust-perceptible differences.1This guide is under the jurisdiction of ASTM Committee E12 on Color andAppearance and is the direct responsi

14、bility of Subcommittee E12.11 on VisualMethods.Current edition approved Dec. 1, 2003. Published December 2003. Originallyapproved in 1996. Last previous edition approved in 1996 as E 1808 96.2For referenced ASTM standards, visit the ASTM website, www.astm.org, orcontact ASTM Customer Service at serv

15、iceastm.org. For Annual Book of ASTMStandards volume information, refer to the standards Document Summary page onthe ASTM website.1Copyright ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States.3.3.5 nominal scale, nscale in which items are scaledsi

16、mply by name.3.3.5.1 DiscussionOnly naming can be performed withnominal-scale data.3.3.6 ordinal scale, na scale in which elements are sortedin order based on more or less of a particular attribute.3.3.6.1 DiscussionLogical operations such as greater-than, less-than, or equal-to can be performed wit

17、h ordinal-scaledata.3.3.7 psychometric function, nthe function, typically sig-moidal, relating the probability of detecting a stimulus to thestimulus intensity.3.3.8 psychophysics, nthe study of the functions relatingthe physical measurements of stimuli and the sensations andperceptions the stimuli

18、evoke.3.3.9 ratio scale, na scale which, in addition to theproperties of other scales, has a meaningfully defined zeropoint.3.3.9.1 DiscussionIn addition to the logical operationsperformable with other types of data, multiplication anddivision can be performed with ratio-scale data.3.3.10 scale, na

19、defined arrangement of the elements of aset of stimuli or responses.4. Summary of Guide4.1 This guide provides an overview of experimental designand data analysis techniques for visual experiments. Carefullyconducted visual experiments allow accurate quantitativeevaluation of perceptual phenomena th

20、at are often thought ofas being completely subjective. Such results can be of immensevalue in a wide variety of fields, including the formulation ofcolored materials and the evaluation of the perceived quality ofproducts.4.2 This guide includes a review of issues regarding thechoice and design of vi

21、ewing environments, an overview ofvarious classes of visual experiments, and a review of experi-mental techniques for threshold, matching, and scaling experi-ments. It also reviews data reduction and analysis procedures.Three different threshold and matching techniques are ex-plained, the methods of

22、 adjustment, limits, and constantstimuli. Perceptual scaling techniques reviewed include rank-ing, graphical rating, category scaling, paired comparisons,triadic combinations, partitioning, and magnitude estimation orproduction. Brief descriptions and examples, along with refer-ences to more detaile

23、d literature, are given on the appropriatetypes of data analysis for each experimental technique.4.3 For reviews of topics in other than visual sensory testingwithin ASTM, see Refs (1, 2).35. Viewing Conditions5.1 Light SourceThe illumination of the specimens inscaling experiments must be reproducib

24、le over the course ofthe experiments. To achieve this, it is essential to control boththe spectral character and the amount of illumination closely inboth space and time. Failure to accomplish this can seriouslyundermine the integrity of the experiments. The spectral powerdistribution of the illumin

25、ation should be known or, if this isnot possible, the light source should be identified as to type andmanufacturer. Information such as daylight-corrected fluores-cent light, warm-white fluorescent light, daylight-filtered in-candescent light, incandescent light, etc., together with param-eters such

26、 as correlated color temperature and color renderingindex, if available, should be noted in the report of theexperiment.5.2 Viewing GeometryAlmost all specimens exhibit somedegree of gonioapparent or goniochromatic variation; thereforethe illuminating and viewing angles must be controlled andspecifi

27、ed. This is particularly important in the study of speci-mens exhibiting gloss variations, textiles showing directional-ity, or gonioapparent (containing metallic or pearlescent pig-ments) or retroreflective specimens, among others. This controland specification can range from correct positioning of

28、 thesource and observer and the elimination of any secondary lightsources visible in the specimens, for the judgment of glossspecimens at and near the specular angle, to more elaborateprocedures specifying a range of angles and aperture angles ofillumination and viewing for gonioapparent and retrore

29、flectivespecimens. When fluorescent specimens are studied, the spec-tral power distribution of the source must closely match that ofa designated standard source.5.3 Surround and Ambient FieldFor critical visual scalingwork, the surround, the portion of the visual field immediatelysurrounding the spe

30、cimens, should have a color similar to thatof the specimens. The ambient field, the field of view when theobserver glances away from the specimens, should have aneutral color (Munsell Chroma less than 0.2) and a MunsellValue of N6 to N7 (luminous reflectance 29 to 42); see PracticeD 1729).5.4 Observ

31、ersGuide E 1499 describes the selection,evaluation, and training of observers for visual scaling work.Of particular importance is the testing of the observers colorvision and their color discrimination for normality. Colorvision tests for this purpose are described in Guide E 1499.6. Categories of V

32、isual Experiments6.1 Visual experiments tend to fall into two broad classes:(1) threshold and matching experiments designed to measurevisual sensitivity to small changes in stimuli (or perceptualequality), and (2) scaling experiments intended to generate apsychophysical relationship between the perc

33、eptual and physi-cal magnitudes of a stimulus. It is critical to determine firstwhich class of experiment is appropriate for a given applica-tion.6.1.1 Threshold and Matching Experiments Thresholdexperiments are designed to determine the just-perceptibledifference in a stimulus, or JPD. Threshold te

34、chniques are usedto measure the observers sensitivity to a given stimulus.Absolute thresholds are defined as the JPD for a change fromno stimulus, while difference thresholds represent the JPDfrom a particular stimulus level greater than zero. The stimulusfrom which a difference threshold is measure

35、d is known as ananchor stimulus. Often, thresholds are measured with respect tothe difference between two stimuli. In such cases, the differ-ence of a pair of stimuli is compared to the difference in an3The boldface numbers in parentheses refer to a list of references at the end ofthis guide.E 1808

36、96 (2003)2anchor pair. Absolute thresholds are reported in terms of thephysical units used to measure the stimulus, for example, abrightness threshold might be measured in luminance units ofcandelas per square metre. Sensitivity is measured as theinverse of the threshold, since a low threshold impli

37、es highsensitivity. Threshold techniques are useful for defining visualtolerances, such as color-difference tolerances. Matching tech-niques are similar, except that the goal is to determine whentwo stimuli are not perceptibly different. Measures of thevariability in matching can be used to estimate

38、 thresholds.Matching experiments provided the basis for CIE colorimetrythrough the metameric matches used to derive the color-matching functions of the CIE standard observers.6.1.2 Scaling ExperimentsScaling experiments are in-tended to derive relationships between perceptual magnitudesand physical

39、magnitudes of stimuli. Several decisions must bemade, depending on the type and dimensionality of the scalerequired. It is important to identify the type of scale requiredand decide on the scaling method to be used before any scalingdata are collected. This seems to be an obvious point, but in theru

40、sh to acquire data it is often overlooked, and later it may befound that the data obtained do not yield the answer required orcannot be used to perform desired mathematical operations.See Refs (3, 4) for further details. Scales are classified into thefollowing four classes:6.1.2.1 Nominal ScalesNomi

41、nal scales are relativelytrivial in that they scale items simply by name. For color, anominal scale might consist of reds, yellows, greens, blues, andneutrals. Scaling in this case would simply require decidingwhich color belonged in which category. Only naming can beperformed with nominal data.6.1.

42、2.2 Ordinal ScalesOrdinal scales are scales in whichelements are sorted in ascending or descending order based onmore or less of a particular attribute. A box of multicoloredcrayons could be sorted by hue, and then in each hue family,say red, the crayons could be sorted from the lightest to thedarke

43、st. In a box of crayons the colors are not evenly spaced,so one might have, for example, three dark, one medium, andtwo light reds. If these colors were numbered from one to sixin increasing lightness, an ordinal scale would be created. Notethat there is no information on such a scale as to the magn

44、itudeof difference from one of the reds to another, and it is clear thatthey are not evenly spaced. For an ordinal scale, it is sufficientthat the specimens be arranged in increasing or decreasingamounts of an attribute. The spacing between specimens can belarge or small and can change up and down t

45、he scale. Logicaloperations such as greater-than, less-than, or equal-to can beperformed with ordinal-scale data.6.1.2.3 Interval ScalesInterval scales have equal inter-vals. On an interval scale, if a pair of specimens were separatedby two units, and a second pair at some other point on the scalewe

46、re also separated by two units, the differences between thepair members would appear equal. However, there is nomeaningful zero point on an interval scale. A common exampleof an interval scale is the Celsius temperature scale. In additionto the mathematical operations listed for nominal and ordinals

47、cales, addition and subtraction can be performed withinterval-scale data.6.1.2.4 Ratio ScalesRatio scales have all the properties ofthe above scales plus a meaningfully defined zero point. Thusit is possible to equate ratios of numbers meaningfully with aratio scale. Ratio scales are often impossibl

48、e to obtain in visualwork. An example of a ratio scale is the absolute, or Kelvin,temperature scale. All of the mathematical operations that canbe performed on interval-scale data can also be performed onratio-scale data, and in addition, multiplication and divisioncan be performed.7. Threshold and

49、Matching Methods7.1 Several basic types of threshold experiments are pre-sented in this section in order of increasing complexity ofdesign and utility of the data generated. Many modifications ofthese techniques have been developed for specific applications.Experimenters should strive to design an experiment thatremoves as much control of the results from the observers aspossible, thus minimizing the influence of variable observerjudgment criteria. Generally, this comes at the cost of imple-menting a more complicated experimental procedure.7.1.1 Metho