aiaa-2002-0532 dot projection photogrammetric technique for shape measurements … ·...

AIAA-2002-0532

Dot Projection Photogrammetric Technique forShape Measurements of Aerospace Test Articles

T. W. Jones and R. S. Pappa

NASA Langley Research Center

Hampton, VA

40th AIAA Applied AerodynamicsConference

14-17 January 2002 / Reno, NV

For permission to copy or republish, contact the American Institute of Aeronautics and Astronautics1801 Alexander Bell Drive, Suite 500, Virginia 20191-4344

https://ntrs.nasa.gov/search.jsp?R=20020021240 2020-04-16T04:34:49+00:00Z

AIAA-2002-0532

Dot Projection Photogrammetric Technique for Shape Measurementsof Aerospace Test Articles

Thomas W. Jones " and Richard S. Pappa

NASA Langley Research CenterHampton, VA 23681-2199

Abstract

Results from initial laboratory investigations with thedot projection photogrammelric technique are

presented for three wind-tunnel test articles with arange of surface scattering and reflection properties.These test articles are a semispan model and a micro

air vehicle with a latex wing that arc both diffusely

reflecting, and a highly polished specularly reflectingmodel used for high Reynolds number testing.

Results using both white light and laser illumination

arc presented. Some of the advantages andlimitations of the dol projection technique arcdiscussed. Although a desirable final outcome of thisresearch effort is the characterization of dynamic

behavior, only static laboratory results are presented

in this preliminary effort.

Introduction

Several image-based, non-contacting videomeasurement techniques, including Proiection Moir6

Interferometry (PMI) and Videogrammetry have been

used for shape characterization in wind tunnelapplications at NASA and elsewhere I. These

techniques are capable of providing quantitativemeasurements of model deformation and dynamics,but each has some limitations or constraints. The

often extensive processing requirement for classicPMI restricts its use in real-time applications.

* Research Engineer-t Senior Research Engineer, Member A1AACopyright © 2001 by the American Institute of Aeronauticsand Astronautics, Inc. No copyright is asserted in theUnited States under Title 17, U.S. Code. The U.S.Government has a royalty-free license to exercise all rightsunder the copyright claimed herein for Governmentalpurposes. All other rights are reserved by the copyrightowner

Videogrammetry, while capable of providing spatialpositions of targets in near real-time at higher

accuracy than PMI, has traditionally relicd on alimited number of either painted or relro-reflcctive

targets attached to the surface of a model. Thislimited number of targets may provide incompletesurface characterization and, in addition, requires

time for target application.

Dot projection techniques attempt to fill the gapbetween the two aforementioned systems, providing a

completely non-contacting spatially complete surfacemeasurement technique suitablc for near real-time

applications. The method of obtaining object spacecoordinatcs described here is based on the projection

of a dense dot pattern (targets). Images arc collected

and processed from two instrumentation grade videocameras to yield the three-dimensional shape of the

object based on the spatial coordinates of theprojected dots. While sevcral systems employingretro-reflective and painted targets and projected

Moir6 fringes have been widely used in wind tunnelsat NASA, dot projection techniques have not yet beenused at all.

The projection of structured light onto a surface hasbeen extensively used for the surface reconstruction

of static objects for more than a decade in variousresearch and industrial applications. More recently

efforts to develop non-contacting optical techniqueshave expanded to include not only modeldeformation measurements during wind tunnel tests,

but also ground based testing of a new generation ofUltra-Light Inflatable Space Structures (UIS) 2'3'4.

Concurrently, the aeronautics community is focusingtechnology development toward a new generation of

biologically inspired flight systems (BIFS) and microair vehicles (MAV) -_. One common aspect these

technology development programs share is theirfocus on membrane construction for critical

components. An example of each is shown in FigureI. The material density requirements for each of these

programs will restrict the use of attached retro-

1

American Institute of Aeronautics and Astronautics

reflective targets and therefore there is a needfor a comparably accurate system with nearreal-time capacity to support dynamic structural

Figure 1: UIS and MAV Vehicle Examples

The potential for commercialization of a dot-projection photogrammetric system has alsobeen recognized by industry. Both GeodeticSystems,Inc. and ShapeQuest, Inc. havecommercially available products based on theconcept of dot projection. 6'14

As technology advancements continue at a rapidpace in high-resolution CCD video cameras,storage capacity, and central processingspeeds, the development of numerous digitalimage processing software packages hasautomated many of the tasks that previouslyrequired a detailed understanding of theprinciples of photogrammetry and imageprocessing, Thus a range of high-accuracyapplications are feasible, even for the novice.For our purposes we seek to provide acontinuous spatial model of the test articlewithout the requirement for attached targets andlengthy post-processing requirements. The goalof this initial investigation was to evaluate theadvantages and constraints associated with theuse of dot projection as a method of gatheringthree-dimensional shape data.

Technology SpectrumThe basic data for photogrammetricmeasurements are images. In general, animage is the result of a perspective projection ofa three-dimensional (3-D) object to twodimensions (2-D). The photogrammetricsolution provides a quantitative relationshipbetween 3-D position, and/or deformation andthe 2-D image plane data recorded by one ormore cameras. Over the last two decades avariety of photogrammetric techniques havebeen used in a number of wind tunnel tests toprovide wing twist and model deformationmeasurements7 Originally, the techniques weredeveloped for use with still-frame film cameras,

characterization, validation, and control of thesemembrane structures.

but the rapid development of CCD technologyled to the successful migration from film to videotechnology. Photogrammetry combined withimage processing has been found to yieldaccurate and near real-time data, but at a limitednumber of points on a given model.

At the other end of the technology spectrum isthe PMI technique, with generally loweraccuracy than the photogrammetric techniques,but with more complete spatial coverage. ThePMI technique projects a grid of equally spacedparallel lines onto the test article. The basicprinciple of the method is that any heightdifference from a reference plane causes a shiftin the projection line of light in the image planeof a recording camera. 8 While the techniqueprovides full-field measurement it is not wellsuited for real-time applications. There areseveral factors that can affect measurementaccuracy for the PMI technique; among themare projected grid pitch, field of view, opticalmodulation transfer function, and illumination.

The dot projection technique proposed here willseek to find a "middle of the road" in the opticalmeasurement technology spectrum, betweenthe superior accuracy associated with the use ofphotogrammetric principles and retro-reflectivetargets (as they have been developed in currentvideogrammetric systems) and the spatiallycontinuous measurements provided with thePMI technique. The basic goal of dot projectionis to apply the same photogrammetric principlesused in the aforementioned videogrammetrytechniques, but without the use of attachedtargets. The dot projection technique aspresented here is based upon two videocameras, but photogrammetric techniques withmore than two cameras would be required tocapture extreme 3-D motion where targets maybecome occluded occasionally. 9

Experimental Discussion

To evaluate the accuracy of white light projectedtargets a laboratory expe'rimen(was designed_todirectly compare the three-dimensionalphotogrammetric accuracy of projected targetsagainst the adopted standard provided byretroreflective targets. Figure 2 shows thelaboratory configuration. The hardwareconfiguration included two Hatachi CCD(640x480) progressive scan video cameras with

2


(

Sony 25-mm focal length lenses, and a NewportMP-1000 slide projector. The test article for this

experiment was a mock-up of the 16 percentscale F18-E/F planform Smart Wing developed

by Northrop Grumman and tested at the NASALangley Transonic Dynamics Tunnel in March1999. l°

,i i!i ii ¸¸Ill!tin _" _ _:_. _'_L_iC__:_ !C.O_ •

Projector _ _ ii

Video Camera :::.....

Figure 2: Hardware Configuration

Calibration

Perhaps the most critical aspect of anyphotogrammetric measurement technique is thecamera calibration. The camera calibration

determines the camera orientation parameters

that relate the 2D image plane and 3D object

space. The process is developed with a set ofcollinearity equations. _ three-tier calibration

target plate, shown in Fig. 3, provided a three-dimensional known target field. The coordinates

of the target field, which were independentlymeasured with a 3-D coordinate measuring

machine, provide an input for the cameracalibration process.

X

¥

11111 m

Figure 3: 3-Step Calibration Target Plate

The calibration process establishes a coordinate

system in which all measured points arereferenced. We aligned the axis of motion

defined by a translation table with the Z-axis as

established by the calibration plate. For thisconfiguration, the Z-axis represents motiontoward the cameras. Once the calibration has

been completed, any camera movement willresult in erroneous data. Similarly, any vibration

or motion of the projector once the initialreference is established will prevent accurate

............ in image sequences.

Projection Hardware

A white-light projector used for the dot projectiontechnique does not have to be calibrated, as isthe case with the PMI technique. The projected

pattern is imaged simultaneously by two CCDcameras. As with most two-camera techniques

an angle of around 90 degrees between theoptical axes of the two cameras is desirable to

maintain precision in the approximate directionthat the cameras are pointed.

A series of slides with different dot sizes were

produced, from which the best projection patternwas selected depending on the object surface

properties and the desired resolution. The slidesused stainless steel screens that were photo-

chemically machined using a photochemicaletching process. _2 The material is commerciallyavailable in sheet-form. Because they are

photochemically etched, the metal is notmechanically perforated; resulting is greatermaterial integrity, and a cleaner, burr-free

perforation. Using the metal screen as a

projection medium produces maximum contrastin the pattern as well as high thermal stabilitycompared to a film slide. The slides were cut tofit a standard slide projector. For the results

presented here a Newport MP-1000 Moir@

projector fitted with a 50 mm Nikon lens wasused in place of a standard slide projector.

Experimental ConfigurationTo generate out-of-plane motion of the wing in

this experiment, it was attached to a Velmex 12inch linear translation table. The table provided

measurement accuracy of 0.001 inch as verified

with electronic dial gauges accurate to 0.0001inch. The projector axis was aligned with thetranslation axis of the table to ensure the wing

remained in the field of projection throughout the

range of motion. A perspective of the wing underdot projection is shown in Figure 4.

3


Figure 4: Test Article with Dot Projection(Simulated Image)

Four retro-reflective targets were positioned attwo spanwise locations along the wing. Thiscomparative analysis used 38 of the projectedtargets. The tracked retro targets are identifiedin green (most extreme targets) and theprojected targets in white (middle set of targets)in Figure 4. The wing translated over a ten-inchrange with image sequences acquired at oneinch increments. The axis of motion was alongthe Z-axis as defined by the camera calibrationprocess.

ResultsAs expected, the retro-reflective targets trackedthe motion of the wing with high accuracy andvery small scatter over the measured range.Using the retro data to check for possible errorin the alignment of the photogrammetric axis (Z)and the axis of translation identified a slightmisalignment of 0.42 degrees, tt was determinedthis small misalignment would not significantlyaffect the data over the measured range. Figure5 highlights the delta change for the retro targetsover the measured range and identifies a meanof 0.9997 inches with a standard deviation of0.0048 inches. The standard deviation for boththe X and Y-axis was approximately 0.001. Thebetter precision in the XY plane was due to thesmall convergence angle between the twocameras that led to reduced precision in the Z-axis.

Point to PoLar Difference: Measured Travel

w/Retro Target

_0980

1 2 3 4 5 fi 7 8 9 l0

Tr'atshfioma][_¢*_m_at(Inches)

Figure 5: Retro Target Incremental Change

Figure 6 shows the scatter in the Z componentbetween the two targeting methods over themeasured range.

15

0.5

.05 L0

°'°_t

)i °2t0

Retros &Projected

i

2

O

i i

4 6

AZure, inch

°°°to Retros ]o Projected Jl

8 10

D

,Jil? &__ __

2 4 G 8 10

AZs_a_,Inck

" o ¢ k o@ o 8 o #o R J;;

0 0 C. _

0

4 6 8 10

AZstaae, inch

Figure 6: Comparison Retro vs. Projected

A closer inspection of the resuits indicates asubstantial variation in the XY plane for theprojected targets while the attached retro targetsshow only small variations in that plane over thesame range of motion. Figure 7 highlights therecorded motion of the projected targets overthe measured range. With the motion of themodel along the Z-axis the projected targetsappear to move closer together. As would beexpected the motion of the projected patternappears to converge toward a point of symmetryfor the projection. This convergence is seen tobe a point in the upper left quadrant of the plot.

4


6

4

2

5

-2

.4

..6

Target Variation Over Measured Range

Retros I[] ,'r°.ie°ted | t t %

i,-.u 0_c

-2

i I i

2 4 6

:; Retros 1Projected

' ' +_.-' ' 0 '8 10 12 14 16 48

X, inch

-8

0 2 4 6 8 40 12

X, inch

Figure 7: Retro/Projected Motion

14 16 48

in XY Plane

To interpret the true surface motion as definedby the projected targets, a fixed reference pointwould need to be included in the data set. Therepeating dot pattern of projected targets can beexported to a CAD environment to generatesplines along various regions of interest. Anobvious shortcoming of projected targets is theinability to provide time histories of a specificpoint of interest on the test article.

4: 0.2,

.o,t

o

._ o.1]

o+o,f

o

[]

[]

i! m H

Residuals from linear least squares fi! (projected only)

[ [] 'P,uj+u+dI

B e _ e B it13

[]

[3

AZs_aoe,inch

[3

[]

O

m H § = u

D

_,Zsta_e,inch

4O

+e 1o

Figure 8: Linear Least Squares Fit (Proj)

Plots of the computed residuals for the projectedtargets for both the X- and Y-axes are shown inFigure 8. One of the targets has a significantlylarger residual variation over the range. Manualinspection of the image sequences identified thetarget in question as an edge target with anincorrect background calculation. Thisdiscrepancy is discussed in the next section.

Discussion - Target RecognitionA factor in achieving maximum accuracy witheither retro-reflective or projected targets is theimage plane diameter of the target. During testplanning, the user should consider thenecessary field of view and the resolution of thecameras. For best centroiding accuracy, thediameter of targets in the camera image planeshould be approximately 5-7 pixels, in the caseof larger structures such as proposed solar sails,this requirement may dictate the need for targetsseveral inches in diameter. A noted benefit ofusing projected targets is the ability to quicklychange projection slides to adjust from larger tosmaller targets when the research focuschanges to an individual component of a largermodel.

Target recognition can be achieved in severalways. Several of the systems currently in use atNASA Langley require initial manualidentification of each target for each camera. Itis obvious that such an effort for a system usinghundreds of closely spaced projected dots wouldbe excessively labor intensive and prone tohuman error. It is therefore desirable that theidentification of targets be automated. A usefulalgorithm for matching the projected dotsbetween two images is based on the epipolar

5


geometry. Epipolar geometry can help establishcorrespondence between dots for each cameraview. As shown in Figure 9 for object point P,the projection centers of the two images, O1 and02, form a plane (the epipolar plane) thatintersects the two image planes. As a result, twoepipolar lines (one for each image) are created.Corresponding points fall on these lines. Oncethe camera positions and orientations aredetermined from calibration, the equations of theepipolar lines are known. 13Since there are manypoints that may fall on this line the approximatedistance from the camera to the test article isused as an additional constraint for the

algorithm.

Z,i

1 I i--'pP<ir,r i,l_- t p2 1

i::f "_'_<': _,. I | : _ |

if.-"- ----JOl 0 "_

Y

/-# P

Figure 9: Epipolar Geometric Model 14

Several factors may contribute to mistakencorrespondence of targets. Object featuressuch as slots, edges or holes will often cause aprojected dot to appear as two distinct targetsfrom one camera perspective, but appear as anormal ellipse from another. Figure 10 showsthe same region of interest on the model butfrom opposite camera positions. Note that theedge between the wing and the flap createsseveral erroneous targets that must be detectedand discarded by the software during imageprocessing. This type of error is also more likelywhen the separation angle between cameras islarge, however reducing this angle negativelyaffects sensitivity in the Z-axis. Automating theprocess of false target discrimination for largetarget fields is difficult.

Figure 10: Target Recognition Errors

Discussion - Target TrackingThere are three widely adopted techniques usedfor target tracking. They are centroid, edge, andcorrelation tracking. Centroid tracking is abrightness based segmentation technique thatcomputes the centroid for the segmented area.Edge tracking seeks to define the target with anedge segmentation scheme that is followed by acentroid calculation. The correlation techniqueidentifies a template to track from frame toframe. For our experiment a centroid trackingtechnique was applied.

There are a number of methods that have beendeveloped to find the subpixel location of atarget. 15 Among the various techniquesavailable are Binary centroid, GaussianDistribution Fitting, Ellipse Least Squares,Perimeter Average and Squared gray scalecentroid.

With projected dots several issues can affect theaccuracy of the subpixel algorithm. An accuratedetermination of the target background isnecessary. With dot projection, the combinationof a high-density dot pattern and a model thatcovers a wide depth of field can negativelyinfluence the definition of background. In manyapplications the search window for the targetsedge is defined as some factor times the largesttarget diameter found on the object. For our

6


experiment the depth of field of the projected

pattern created targets that ranged from 21 to 45pixels over the full span of the model. When

using a high-density projection pattern, adjacenttargets only a few pixels away may negatively

influence target centroiding when a generalsearch window based on the diameter of the

largest target is used to define the background.With our experiment the 38 projected targetswere selected over a region that minimized their

size variation and thereby avoided this issue of

background definition.

As simulated in this paper using a linear stage to

move the entire wing in the out-of-planedirection, the technique could support a wide

range of dynamic motion. A reconstructed

rendering of the model surface created using allof the projected targets from a single image setof the experiment is shown in Figure 11. With

the use of external control targets (i.e. stationary

background targets) a rigid frame of referencecould be established to reference sequential

renderings of the model.

Figure 11 : Surface Rendering of Wing

The step from capturing static surface contourmeasurements to dynamic motion may requireadditional hardware. One possibility is a

configuration using a pulsed light source as the

projector. The pulsed light output would freezethe motion to prevent blurring of the projecteddots. To ensure the light pulse occurs duringthe camera open shutter period an external

synchronized trigger will be required.

not critical, the projector was positioned 10 feetfrom the test article to ensure sufficient intensity

When considering the use of white-light dot

projection over the more common retro-reflective

targets, consideration must be given to thereflective nature and surface finish of the test

article. Typically wind tunnel models havesurface finishes ranging from highly polished

steel to painted diffuse white panels. Asdemonstrated in Figure 12, the contrast in target

and background is very low for a highly reflectivemodel surface due to the substantial reflection of

light away from the cameras. In cases wherethe specular reflection direction is toward one ofthe cameras, the light source may create bright

spots that prevent target recognition in theimmediate area.

Deformation measurements at the National

Transonic Facility (NTF), where model surfaces

are always highly polished, routinely require thetargets to be painted on the model surface. 16

The polished paint diffuse targets are typically0.0005 inch thick with a surface roughness of

less than 10 pinch. Ordinary lights (non-laser)

can provide sufficient illumination. While thistechnique has been demonstrated to beeffective for deformation measurements at the

facility, the procedure for painting the targets isdetailed and time consuming. Similarly, repair

time for damaged targets can be substantial.

Ambient lighting can also create reflections on apainted target model. Since a laboratory

investigation of the use of projected targets onhighly reflective model surfaces yielded poorresults due to the low contrast and surface

reflections of the source projector, further

developments will be necessary before the dot

projection technique is suitable for highReynolds number facilities such as the NTF.

Model Surface and Lighting

To take advantage of the full dynamic rangeavailable with a standard 8-bit digital video

camera the settings should be optimized withthe combination of shutter speed and aperture to

remove extraneous light. Due to the variationsin ambient lighting in many wind tunnel facilities

a higher power projector may be necessary toensure adequate contrast. Since the distancebetween the retro targets and the light source is

7


Figure 12: Highly Polished Metal Surface

Laser Projected TargetsThere are advantages and disadvantages tousing a laser as a replacement source forprojecting the dot pattern. A suitable laser willprovide a concentrated diffraction limited beamthat is generally visible above the ambientlighting in most environments. This highintensity beam may be able to providecomparable signal-to-noise ratios commonlyfound with the use of retro targets. In addition, alaser could also be chosen that would operate inthe near infrared. A pulsed laser diode emittingin the near infrared similar to that used in theaforementioned PMI systems, coupled withsuitable wavelength filtering, may supplysufficient contrast to allow lights-on operations inmost research facilities.

Among other advantages noted is the narrowspectral bandwidth of the laser, which providesthe user the unique ability to remove most of theextraneous light with an interference filter on thecamera. Diode lasers are another option thatcompare favorably to available white lightprojectors. A diode laser offers a small (fewinches) and self-contained package thatoperates with low power requirements. Whencoupled with a diffractive pattern dot generatorthe combination is routinely used in lasermachining, laser marking, alignment systemsand computer vision. However, note that safetyconsiderations will be a factor when using alaser as the projection light source.

laser beam and optical axis of a camera. TheA.G. Davis index head has an accuracy of betterthan 1 arc second. For this experiment a 5mWHeNe laser was selected. The MAV model wingsection used was made of latex and provided aconsistent contrasting background for targetcomparison. The model was placed 48 inchesfrom the laser and camera station. The cameraoptical axis and the projected beam axis werealigned by sight. The setup is shown in Figure13. A stationary Kodak DCS 460 6 megapixel(3060x2048) camera acquired the imagesequences. The camera was stationary duringthe image sequences. The camera focal lengthand aperture settings remained constant for allimages. The camera electronic shutter wasvaried to obtain approximately the same relativeexposure for each trial. All targets wereapproximately 0.25 inches in size.

Camera calibration yielded an estimatedsensitivity of 0.0055 inches/pixel. Thecalibration scheme employed for the Kodak DCSprofessional series camera was based on aphotogrammetry technique known as "bundleadjustment." The scheme utilizes multiplecamera stations, multiple convergent views, anddifferent camera angles to determine accuratelythe interior orientation parameters. Using_:the-Commercial Photorn-0deie_ software, from EOS

Systems Inc., the process requires the user tomanually identify a series of control pointsoneach of the Calibration images. The bundlesolution then Computes the interior orientationand distortion parameters of the camera.

.........................................JJi¸

Figure 13. Laser Target Experiment Configuration

As part of the evaluation of the centroidingaccuracy using retroreflective, laser, or whitelight projected targets, a micro air vehicle (MAV)test article was mounted on a precision rotaryindex head approximately perpendicular to the

Visual observations indicated that laser specklemight adversely impact centroid determination.Due to the concentration of light for the laser thetarget saturated (255) on an 8-bit grayscale

8


reading. A comparison of the variation in

grayscale readings across the retro target andlaser target shows a more well defined thresholdtransition for the retro target. A graph of the

typical cross sectional intensity for each type of

target is illustrated in Figure 14. The centroidcalculation requires an accurate determination of

the target background for a defined search area.The three graphs on each figure represent theRed-Green-Blue content of the target. As

expected with a HeNe laser, the saturated signal

is mostly red.

_=-LinePiot_g:Laser....................Projro:,ictio"MA. _-MAY__..................

,J Vill i- " ¢ i _L

7_0. _i_:._.-.,_iit_

7 iIia _0

..........L.ineP,gtiI_._..Reti'o - MAV

2__oo71 5°-

I00-

50-

Ol ..... u0 106

Figure 14: Cross Sectional Comparison ofLaser and Retro Targets

The targets were positioned on the test article so

the center target was aligned with the rotationcenter of the index head. The target alignmentwas verified by rotating the object 180 degrees

and verifying the edges were overlapping oneach view. The index head and the laser were

leveled using a precision 3-axis accelerometer

package.

To investigate the effect of laser speckle on thecentroid computation, a series of images weretaken at incremental rotation angles. The

images were shot over a _40 degree range in 5-

degree increments. The calculated mean for thecentroid of each target data set was used as the

reference for comparing the scatter between thetargets. As indicated by the charts in Figure 15,there are much larger variations in the

centroiding accuracy for the laser-projected

target than for the retro target.

Centroid Variation - Retro Target

393.3125

,<;_ 393.312

E 393.3115

393.311

435.56 435.562

Gx = 0.0006

Gy = 0.0003

: '/i

435.564 435.566

Image Plane XAxis

Centroid Variation - Laser Target

I (Sx = 0 755'_ 462 -'_r'_--_ , --_ ..... '

461.8 t ,-lll ....;_"r (52_ = 0'297

461461"4t "'_--:_7--__.2 ID'

_ 460.8 __ 461 ...... _ _ :4!b :i: ;; ;;

826.5 828.5827 827.5 828

Image Plane X Axis

Figure 15: Centroid Variation Laser/Retro

The major disadvantage of laser projection isspeckle. While there are several optical

techniques for minimizing this effect, the errorcontribution when calculating the centroid canbe significant) 7 Note that the coherent nature ofthe HeNe laser and the diffuse nature of the

surface are the primary cause of the speckle

noted in this experiment. It would be expectedthat the effect of speckle on centroid calculation

could be reduced using a multi-mode, lowcoherence length laser diode with limited spatialcoherence. Further investigations are plannedwith alternative laser sources.

Conclusions

Initial experimental investigations wereconducted to evaluate the benefits of dot

projection as an alternative to attached targets

for shape characterization of various types ofmodel surfaces. Several issues have been

discussed with regard to using projected dots

from either white light or laser sources.

Regardless of the source, the projected dotsmust have high contrast to achieve accuratemeasurement. We were able to achieve good

contrast for the projected white light targets byadjusting the aperture at a cost to depth of field.

The depth of field can also create problems in

9

American Inslilule of Aeronautics and Astronautics

target centroiding when provisions are not madefor the variable dimension of targets over thefield. Our investigation selected a target patternto avoid this issue. While the experimentalinvestigation was not set in an actual windtunnel, the flexibility associated with the use of aprojector and a variety of projection slidesshould be adequate for any model tested inNASA Langley wind tunnel facilities. While thetechnique seeks to include the strengths ofconventional videogrammetry and PMI, it hasseveral weaknesses regarding target recognitionthat would prevent its rapid acceptance in real-time applications. Targets near an edgegenerally require manual intervention for targetrecognition. Complex model surfaces thatcontain a significant number of lines, breaks,and holes create a higher probability ofinaccurate target matching between images.

on Micro Air Vehicles is planned for the BasicAerodynamic Research Tunnel at NASALangley in February 2002. A parallel effort forshape characterization of a 10m solar sail will beconducted in March 2002 in the 16m VacuumChamber at NASA Langley. Further laboratoryexperimental investigations are planned with theuse of pulsed light projection sources.

AcknowledgementsThe contributions of Mr. AI Burner and Mr. ScottSealey, NASA Langley Research Center, in theexperimental testing of dot projection on thevarious test articles are greatly appreciated.

Overall, the benefits identified using projectedtargets are significant when a measurement taskrequires a completely non-contacting technique.Among the Chief benefits'noted is the ability togenerate full field coverage and to quicklychange projection patterns to meet changingtest requirements. From a productivitystandpoint, the use of dot projection wouldrequire a minimum of preparation time. From adata analysis standpoint, the most significantshortcoming of projected targets is the inabilityto provide time histories of a specific point ofinterest. However, with the use of externalcontrol points the--d-ata Could be exportecl to aCAD environment that supports the generationof 3D models with the use of splines createdfrom the photogrammetric data.

Disadvantages of retro-reflective targets are thatthey must be placed on the test article inprescribed places to describe the motion orposition of the surface under observation andthe requirement for a closely aligned lightsource. Often this process can be timeconsuming or the coverage would be tess thandesired to fully describe the surface motionwithout interpolation. While material propertiesmay preclude the use of attached targets forresearch in various new aerospace programs itis likely that retro targets will remain thestandard for conventional model testing in mostwind tunnel facilities.

==

Future Plans _ _A more detailed investigation of the constraintsassociated with the use of white light projection

l0


References

Burner, A.W.,Flcming, G.A., and Hoppe, J.C.,

"Comparison of Three Optical Methods forMeasuring Model Deformation", AIAA Paper 2000-

0835, Jan. 2000.

_-Pappa, R. S., Jones, T. W., Robson, S., and Shortis,M. R., "Photogrammetry Methodology Development

for Gossamer Spacecraft Structures", Proceedings ofthe 3rd AIAA Gossamer Spacecraft Forum, Denver,

CO, April 2002.

3 Pappa, R. S., Woods-Vedclcr, J. A., Jones, T. W.,

"In-Space Structural Validation Plan for a Stretched-Lens Solar Array Flight Experiment", To be

presented at the 20th International Modal AnalysisConference, Los Angeles, California; February 4-7,2002.

+ Dharamsi, U.K., Evanckik, D.M., and Blandino,

J.R., "Comparing Photogrammetry with aConventional Photogrammetry with a Conventional

Displacement Measurement Technique on a Square

Kapton Membrane", To be presented at the 43thAIAA Structures, Structural Dynamics and Materials

Conference, Denver, CO, 22-25 April, 2002.

5 G.A.Fleming, et. al, "Projection Moird

Interferometry Measurements of Micro Air Vehicle

Wings," SPIE International Symposium on OpticalScience and Technology, San Diego, CA July 29-

August 3, 2001.

6 Ganci, G. and Brown, J., "Developments in Non-

Contacting Measurements Using Videogrammetry",Boeing Large Scale Metrology Seminar, St. Louis,

MO, February 13-14, 200 I.

7 Burner, A.W., T. Liu, "Videogrammetric ModelDeformation Measurement Technique", Journal of

Aircraft, Volume 38, Number 4, Pages 745-754,2001.

s Pirodda, L., "Shadow and Projection Moird

Techniques for Absolute or Relative Mapping ofSurface shapes", Optical Engineering 21 (4), pp. 640-

649, July/August 1982.

'_Fraser, C.S. and Shao, J., "On the Tracking of

Object Points in a Photogrammetric Motion Capture

System", Videometric and Optical Methods for 3DShape Measurement (SPIEL Vol. 4309, San Jose,

CA, January 22-23, pp. 212-219, 2001.

1(3 r, le -r mmg, G. A., Burner A. W.,"DeformationMeasurements of Smart Aerodynamic Surfaces",SPIE Paper No. 3783-25, Presented at the 44 th

Annual SPIE International Symposium on Optical

Science, Engineering, and Instrumentation, Denver,

CO, July 18-23, 1999.

_ T.Liu, R. Radeztsky, S. Garg and L. Cattafesta, "A

Videogrammctric Model Deformation System",AIAA Paper No. 2001-0560, 39 mAIAA Acrorspace

Sciences Meeting & Exhibit, Reno, NV, January 8-

11,2001.

12BuckBee-Mears Inc., 2001, Micro-Etch© [On-

Line], [http://www.buckbecmears.com].

t3 Maas, H.-G., "Automated photogrammetric

Surface Reconstruction with Structured Light",Industrial Vision Metrology (SPIE), Vol. 1526,

Winnipeg, Manitoba, pp. 70-77, 1991.

14ShapeQuest Inc., 2001, ShapeCapture©[On-Linc],

[http://www.shapequest.com].

t._Shortis, M.R., Clarke, T.A., and Short T., "A

Comparison of Some Techniques for the SubpixelLocation of Discrete Target Images", Videometrics

III (SPIEL Vol. 2350, Boston, MA, 1994.

_6Burner, A.W., and Martinson, S.D., "Automated

Wingtwist and Bending Measurements UnderAerodynamic Load", AIAA Paper 96-2253, NewOrleans, LA, June 17-20, 1996.

_7Clarke, T.A., "An Analysis of the Properties of

Targets Used in Digital Close RangePhotogrammctric Measurement", Videomelrics III(SPIEL Vol. 2350, Boston, MA, 1994, pp. 251-262.

11


aiaa-2002-0532 dot projection photogrammetric technique for shape measurements … ·...

Documents