Stereophotogrammetry as anAnthropometric Tool*

H. T. E. HERTZBERG,

C. W. DUPERTUIS, AND I. EMANUEL

ABSTRACT: This paper briefly reviews previous biological applicationsof stereophotogrammetry, and outlines with illustrations the present pro­cedures used to draw human body contours at liZ-inch intervals. It com­pares the dimensions derived from plolted profiles with those taken byhand on the subjects themselves. It discusses the utility of stereo datafor special anthropometric purposes, and mentions further applicationsfor other biological sciences.

F OR centuries men have measured anddrawn maps of the earth they live on,

but only recently have they begun to seekprecise delineation of their own bodies.This paper describes a method of accu­rately mapping the dimensions and curva­tures of the body, and of establishingwhere and how much the surface of thebody protrudes from a known plane of re­ference.

This work supports the Air Force effortto design form-fitting, protective garmentsthat insure pilot survival at high altitudes.To design a protective garment that fitssnugly to a man's skin, one must know theform of that skin. While conventionalanthropometry is useful for the sizing ofcockpits, personal equipment, loose cloth­ing and the like, it is not efficient for sizingskin-fitting garments. For this special pur­pose, one is forced to use the techniques ofstereophotogrammetry similar to thoseemployed in the mapping of land-forms byaerial photography.

The basic principle of stereophotogram­metry is exactly that of binocular vision.The two eyes send slightly different imagesof an object to the brain, where they areinterpreted in terms of depth as well as oflength and breadth. Similarly, if two bino­cular or stereophotographs of an objectare juxtaposed so that the left eye sees theleft photograph and the right eye sees theright photograph in proper relation, 1 heperception of depth can be as clear as if theobject were seen directly. By suitableoptical devices, stereophotographs ofnearly any size can be made to give theeffect of relief.

These stereoscopic facts have long been

known and applied in cartography. Theirusefulness in the medical and dental fieldswas pointed out in the early 1920's byMannsbach and others. Zeller (1939) pub­lished actual contours of a man's face;Thalmann-Degen (1944) used contours toindicate the degree of swelling of a jaw;Lacmann (1950), apparently to emphasizethe universal utili ty of stereo, presen tedcontours not only of a man's face but alsoof the rear end of a cow; Berner (1954)made calculations of the surface area of ahuman body by stereophotogrammetry;Leydolph (1954) published contours ofcows and sheep taken years previously,and discussed the uses of stereo in animalhusbandry; and Bjorn, Lundquist andHjelmstrom (1954) assessed the progress offacial swellings for dental purposes. Afterwe had written our own paper, we learnedthat Miskin (1956) had applied stereotechniques to the study of new-born in­fants, including body volumes and surfaceareas, as well as the assessment of bodybuilds and local deformations. The use ofstereo-contouring methods in sculpture hasbeen reported by Saralegui (1954) andEnglemann (1956).

All these studies were made with stereocamera equipment manufactured especi­ally for such applications. We have usedonly standard Air Force cameras modifiedfor our purposes, and so far as we know,our subjects are the first to have been fullyand accurately contoured over virtuallytheir entire bodies.!

1 This point is not easily determined becausemany articles are in European journals whichare difficult to obtain (continued on next page)

* Presented at meeting of Ohio Region of the Society at Columbus, Ohio, October 11, 1947.

942

STEREOPHOTOGRAMMETRY AS AN ANTHROPOMETRIC TOOL 943

ZERO PLANESTEREO LIGHT

CAME. A \

\ 11 j't,I 3'-6' l2( j (J~

~----+-l-~-H--- \ ~~ ) J4"o"/ I ).."1 'j-L/II ,L....~l.

STEREO I LIGHT

CAMERA 12'.0"

SOMATOTYPE

CAMERA

17'-0·

ALL CAMERAS;"-," ABOVE FL.OOR

FIG. 1. Geometry of Experimental Stereo Equipment.

The equipment for our experiments fol­lows conventional photogrammetric prac­tice by maintaining a carefully-alignedsolid geometry based on fundamentalstereoscopic principles.2 A vertical plane isestablished by means of an aluminum armaffixed at right angles to a wall about sevenfeet above the floor. This datum plane ismarked by 3 X 5-inch cards correctly posi­tioned within five one-hundreths of aninch. In this plane is the center-line of theturn-table on which the subject stands.Two K-17 aerial cameras, using 9!-inch by9!-inch film, are mounted 3! feet above thefloor, five feet apart and twelve feet fromthe datu m plane, their optical axes parallel.The camera shutters have been modifiedfor synchronization with electronic flashlamps and their 90 degree metrogon lenseshave been focussed on the datum plane.3

For certain technical reasons a standard4X5 view camera also is mounted 17 feet

in the United States. Actually much of ourbibliography came in on inter-librarialloan afterwe had developed our techniques.

2 We are indebted to Capt. L. E. Ross, for­merly of WADC, now of Rome Air Develop­ment Center, who kindly worked out severalgeometries for us to try. The one described is theone we found most effective for our purposes.

3 We are grateful for the skill and ingenuity ofall personnel at Technical Photo Division,WADC, who processed our film and advised onphotographic matters. Especial thanks are dueto Mr. James E. Taylor and Mr. K. J. Jones ofthe Photo Instrumentation Branch, and theirassistants, Mr. Elmer Fisher and A/2C HaroldE. Geltmacher, who modified our cameras andmade them work.

from the zero plane, shooting between thetwo K-I7's. The over-all geometry is shownin Figure 1. The K-17 is shown in Figure 2.

Before deciding to standardize our workon the geometry and positioning of thecameras just described, two other align­ments of the equipment were tried out inthe initial stages of the research. I n one ofthese orientations the cameras were placedfive feet apart with their optical axes per­pendicular to the datum plane, with eachlens six feet from the datum plane. In tneother alignment the cameras were also fivefeet apart, but were pointed directly at thesubject standing in the center of rotationof the turntable and six feet from him. Thisorientation gave an angle of 40 degreesbetween camera axes. Both of these geo­metries were soon abandoned, however, asat this short shooting distance neithercamera could pick up satisfactorily thecontrol points on both sides of the turn­table, thereby making it virtually impossi­ble to achieve absolute orientation on theKelsh Plotter. Furthermore, because of thelarge ratio of the object heigh t (thicknessof the subject's body) to the object dis­tance, it was found that the vertical move­ment required of the floating dot far ex­ceeded that provided by the standardKelsh tracing table.

During the photographi'c procedure thesubject stands on the turntable facing thecameras so that his lateral ve~tical planelies as nearly as possible in the datumplane of the system. We have found thatfor our purposes his legs must be somewhatspread and his arms outstretched. To pre­vent movement during photography, the

944 PHOTOGRAMMETRIC ENGINEERING

FIG. 2. K-17 Camera (Modified) on Tripod.Two of these cameras were used to obtain thestereo pairs.

subject rests his arms on two arm supportson the turntable. Before photography thesubject must be suitably marked. Thesemarks are of two kinds: (1) marks for bodymeasurement, to permit comparison of di­rect anthropometric data with' the data de­rived from the photographs; and (2) stip­pling. The measurement marks are usuallycrosses ringed by a circle; located at certainpoin ts on the body. The stippling is neces­sary because the eye cannot easily discerndepth when there is no contrast from onearea to another. 4

If unlimited space and equipment had

4 We are indebted to Mr. L. ]. Kosofsky, for­merly of Wright-Patterson Air Force Base, forthe suggestion to employ stippling. It has solvedthe problem perfectly. Mr. Kosofsky also cali­brated the lenses for the 12-foot focus.

FIG. 3. The Kelsh Type StereoscopicPlotter. Model 5010.

been available it would have been ideal toarrange a ring of cameras and of lightsources around the subject to photographhim in rapid sequence, thereby virtuallyeli mi na ti ng his move men t. \Ne did nothave ideal conditions, so were forced to thefollowing technique. When properly posedthe subject was photographed simultan­eously by all three cameras and then wasrotated 90 degrees and photographedagain. This was repeated each 90 degrees.He made every effort not to move betweenphotographs, as this would have intro­duced error by varying his protrusion fromthe zero plane.

vVhen the negatives were developed,diapositives were prepared for use in theKelsh plotter5 (Figure 3). By means of thismachine, contour maps of the human bodywere drawn for front, back, and side eleva­tions. 5

.G After much experimentation it was

5 We thank the U. S. Army Research and De­velopment Laboratory, Corps of Engineers, forkindly making their Kelsh Plotter available.

Mr. E. D. Sewell, Chief, Aerial PhotographicBranch, and Mr. C. E. Berndsen, PantographicConsultant, were most helpful in teaching usthe principles and practice of contour plotting.Mr. Robert Shearer, formerly of WADC, nowof Rome Air Development Center, also assistedgreatly to teach us the techniques of handlingthe Kelsh Plotter.

6 All the contour maps shown in this reportwere produced by C. W. Dupertuis.

STEREOPHOTOGRAMMETRY AS AN ANTHROPOMETRIC TOOL 945

CJ

~I

D

CI

J~r-I''=;=::-~4,Ti')ei~"=-=-:):;-:J~-,,~_ '; ~ ~ .. ,---L,o". .

FIG. 4. Front View of Subject and Corresponding Contour Map, showing subject's positionon turntable, measurement marks, stippling, and cards indicating the zero datum plane. Contoursare at t-inch intervals. Original drawn to about 1:4 scale.

found that a scale of 1:4 and contour in­tervals of one-half inch were suitable forour purposes.

Once the contour maps are drawn it ispossible to plot profiles or cross sections be­tween any two known points on the map,as is done daily in geological science. Oneof our immediate purposes, of course, isthe extraction of body dimensions; ourlong-range purpose is a more precise de­scription of body types in statistical terms.Figure 4 shows the subject's position onthe turntable, the marks and stippling, andthe front and back views contoured at one­half inch intervals. The several 3 X 5 cards

visible on both sides of the subject andover his head indicate the zero referenceplane. The circled crosses on the subject'sskin are the points used for measurement;the lines from the shoulders to the feetdelineate four experimental surfaces­front, each side and back. 7 These lines neednot be precisely drawn, as they would varywith each person.

Although the contour maps constitutethe basic data of this study, by themselves

7 We are indebted to A/2C James 1'. Barter,formerly of the Anthropology Section, for thissuggestion. Airman Barter also assisted in the'design of the turntable.

OJ

I_I

FIG. 5. Rear View of Subject and Corresponding Contour Map.

946 PHOTOGRAMMETRIC ENGINEERING

TABLE I

COMPARISON OF ANTHROPOMETRIC AND PHOTOGRAMMETRIC MEASUREMENTS

Dimension

Upper Chest ArcI ntra-Nipple ArcLower Abdomen ArcUpper Shoulder ArcTrunk-Shoulder ArcCircumference at 1 ipplesCircumference at WaistSuprasternale Height

From Actual Body

9.65 Inches8.30 Inches

12 . 50 Inches10.50 Inches39.50 Inches37.7 Inches37.7 Inches57.1 Inches

From Contour Map

9.58 Inches8.24 Inches

12 .40 Inches10.35 Inches39.51 Inches37.3 Inches37.2 Inches57.0 Inches

they are not enough. The next step is theplotting of profiles. These can be surfacearcs, as between the nipples, or completecircu mferences at any elevation. To plotthem one must first establish a line be­tween two known points on the contourmap, then measure the distances from onecontour to the next as they intersect thatline. After each distance is brought to fullsize by applying the known scale of thecontour map, it is plotted on cross sectionpaper with due regard for the difference inelevation between contours. Figure 5shows four circumferential profiles plottedisometrically for the planes A, B, C and D,that are shown as horizontal lines on thecontour maps. These planes, however, neednot always be horizontal. They may bedrawn in any direction across the contours,permitting the plotting of the surface tracealong that plane.

A comparison of certain anthropometricmeasurements taken on the subject andphotogrammetric measurements derivedfrom the contour maps are presented inTable 1.

It can be seen from Table I that there is

very little disagreement between the twoonly three being over 1/10 of an inch andamounting generally to less than one percent of difference between dimensions.Even the largest, Circumference at Waist,amounts to only 1.25 per cent.

In our opinion, stereoanthropometry ismuch more precise than hand anthro­pometry, and the question is no longer,how accurate is the system, but how ac­curate is the man! For this surface changesevery time he moves, by breathing orotherwise, and it becomes very difficult toreplace him-and to remeasure him!­exactly in the same position. It appearsprobable that the major discrepancies inTable I are due to postural changes by thesubject.

Having described an exacting and some­what expensive system, we think it is onlyfair to assess its benefits in terms of whatcan be done by conventional anthropome­try. Table II compares the capabilities ofthe two systems in providing ten categoriesof data.

From Table II it can be seen that con­ventional anthropometry can give body

TABLE II

COMPARISON OF SYSTEM VALUES

Values

1. Diameters2. Surface Arcs3. Circumferences4. Circumference Shapes5. Body Proportions6. Body Surface Area7. Body Volume8. Record of the White Man9. Expense

10. Simplicity

Anthropometry

YesYesYesNoNoNoNoNoNoYes

Stereophotogrammetry

YesYesYesYesYesYesYesYesYesNo

STEREOPHOTOGRAMMETRY AS AN ANTHlWPOMETRIC TOOL Y47

FIG. 6. Isometric Plot of Profiles at 4 Levelson the Torso.

diameters and surface dimensions whichare adequate for the sizing of loosely-fittingclothing and of work places. Stereophoto­grammetry, on the other hand, is capableof providing not only these data, but alsobody proportions and the form of thesurface. In short, stereophotogrammetry isso far the only means of providing simul­taneously the record, the dimension, andthe form of all the features of the body.Furthermore, from subsequent work wehave found that the measurements of totalbody volume and the volumes of individualbody segments very closely approximatethose calculated by the Dubois formula(Dubois and Dubois, 1916). In addition,we have been able recently to determinethe surface area of various body parts witha high degree of accuracy.

We believe that through use of thestereophotogrammetric method we even­tually will be able to give close approxima­tions to the total body surface area. Cer-

tainly the method should find increasingapplication for the medical and dentalpurposes to which other workers have al­ready put it; and it should become veryuseful among biological sciences for themeasurement of growth and age changesof any animal, including man. It alsoshares the exceedingly valuable featurepossessed in common by all photographicmethods, namely, that a large sample canbe stored in a small space to be restudiedfor new dimensions or purposes at anytime.

LITERATURE CITED

Berner, J., Jr., 1954. "Calculation of the BodySurface Area by Photogramll1etry." TheScandinavian Journal of Clinical and Labora­tory Investigation, Vol. 6, No.4.

Bjorn, H., G. Lundquist, and Petter Hjelm­strom, 1954. "A Photogrammetric Methodof Measuring the Volume of Facial Swell­ings." Journal of Dental Research, June.

Dubois, D., and E. F. Dubois, 1916. "A Height­Weight Formula to Estimate the SurfaceArea of Man." Society for Experimental Bi­ology and Medicine, Vol. 13.

Engelmann, VI/. F., 1956. "PhotogrammetryApplied to Making Portraits." PHOTOGRAM­METRIC ENGINEERING, Vol. XXII, No.2,April.

Lacmann, Otto, 1950. "Die Photogrammetrie inIhrer Anwendung auf Nicht-Topograph­ischen Gebieten." S. Hirzel Verlag, Leipsic.

Leydolph, W. K., 1954. "Stereophotogram­metry in Animal Husbandry." PHOTOGRAM­METRIC ENGINEERING, Vol. XX, No.5, De­cember.

Mannsbach, M., 1922. "Die Stereophotogram­metrie als Hilfsmittel in der Orthodontie."Deutsche ZeitschriJt jiir zahnarztliche Grtho­piidie, p. 105.

Miskin, E. A., 1956. "The Application of Photo­grammetric Techniques to Medical Prob­lems." The Photogrammetric Record, Vol. II,No.8, p. 92, October.

Saralegui, A. M., 1954. "What is Photosculp­ture?" PHOTOGRAMMETRIC ENGINEERING,Vol. XX, No.1, March, 1954.

Thalmann-Degen, Paula, 1944. "Die Stereo­photogrammetrie, ein Diagnostisches Hilfs­mittel in der Kieferorthopadie." DoctoralDissertation, University of Zurich, Switzer­land.

Zeller, Max., Prof. Dr., 1939. "Die Photogram­metrischen Methoden und ihre Anwendung."Section in: "Vermessung-Grundbuch-Kar­te." Festschrift zur schweitzerliche Landes­ausstellung, p. 46.