Utafiti huu unapendekeza mbinu mpya ya kuunganisha sensa ili kupima uga wa kasi na uhamisho wa vipimo vitatu wa uso unaotetemeka kwa kutumia mwelekeo mmoja tu wa kuona wa kioptiki. Mbinu inaunganisha vipimo vinavyokamilishana vya kamera ya kasi kubwa na vibromita ya leza ya kuchanganua ya mhimili mmoja, vilivyowekwa kwenye mhimili uleule wa kioptiki. Algoriti ya optical flow inayotumika kwenye picha za kamera ya kasi kubwa hupima vipengele viwili vya mtetemo ndani ya image plane; vibromita ya leza hupima kipengele cha mtetemo wa out-of-plane kilicho sambamba na mhimili wa kamera. Baada ya data zilizo kwenye coordinate grids tofauti kusawazishwa na kuinterpolateiwa, kasi ya mtetemo yenye vipengele vitatu kwenye uso hupatikana.

Kwa ajili ya kuhesabu uhamisho, particle advection ilitumika kwenye uga wa kasi wa vipimo vitatu uliopimwa. Kwa njia hii, ilihamishwa kutoka Euler approach, ambayo huonyesha kasi katika maeneo yasiyohama angani, kwenda Lagrange approach, ambayo huonyesha trajectories za virtual particles zinazosogea pamoja na uso. Mabadiliko haya ni muhimu hasa kwa kuchunguza mwendo wa nyuso laini zenye mabadiliko makubwa ya umbo, kama vocal folds.

Mbinu ilithibitishwa kwenye synthetic vocal folds za silicone zinazojitetemesha kutokana na airflow. Matokeo ya monoscopic measurement yalilinganishwa na stereoscopic reference method iliyojumuisha kamera mbili za kasi kubwa zilizowekwa kwa angle ya takribani 40 degrees pamoja na three-dimensional tracking ya surface markers 12. Kwa total three-dimensional displacement magnitude katika points kumi na mbili, average difference ilikuwa %5,2. Phase differences zilikuwa kwa wastani %1,1 katika image-plane components na %1,5 katika out-of-plane component.

Displacements katika image plane kwa ujumla zilitoa matokeo yaliyo karibu zaidi. Average image-plane difference ilikuwa %5,4, huku average difference ya out-of-plane component ikiwa %10,0. Katika baadhi ya points, out-of-plane difference ilifikia %21,7. Hii inaonyesha kwamba mbinu haina accuracy sawa katika coordinates zote. Hata hivyo, fundamental vibration frequency na phase relationships katika harmonics za kwanza kwa ujumla zilioana vizuri kati ya mbinu hizo mbili.

Baada ya validation, mbinu ilitumika kwenye mfumo wa synthetic vocal folds uliokuwa umeongezewa vocal tract yenye urefu wa 200 mm, jambo lililopunguza optical access. Katika experiment hii, uso uliwakilishwa na jumla ya virtual particles 4.320; opening, closing na surface-shape deformation motions za vocal folds zilijengwa upya katika vipimo vitatu. Local vibration velocities zilifikia takribani 3 m/s na harmonic components hadi mara 30 ya fundamental frequency zilipimwa.

Fundamental vibration mode iliundwa hasa na motion ya image plane inayofungua na kufunga nafasi kati ya vocal folds. Katika second na third harmonic modes, out-of-plane motion katika direction ya camera axis ilitawala; katika second mode, vibration node ilionekana katikati ya uso. Matokeo haya yanaonyesha kwamba single measurement point inaweza isiwakilishe complex vibration behavior ya uso mzima.

Faida muhimu zaidi ya mbinu ni kwamba haihitaji uso uonekane kutoka directions mbili au tatu tofauti. Hata hivyo, single optical direction haimaanishi single device: mfumo hutumia kamera ya kasi kubwa na laser scanning vibrometer kwa pamoja. Katika implementation ya sasa ya utafiti, camera recording na laser scanning hazikufanywa simultaneously; zilifanywa sequentially na phase-locked kwa common reference signal.

Tathmini kwa mtazamo wa Türkiye: Utafiti haukufanywa Türkiye, kwenye vocal folds halisi za binadamu au katika clinical conditions. Hata hivyo, mbinu inaweza kutathminiwa katika laboratories za fluid mechanics, biomedical engineering, acoustics, phoniatrics na optical measurement nchini Türkiye kwa ajili ya kuchunguza synthetic phonation setups ambazo optical access imewekewa kikomo kwa direction moja. Kabla ya clinical use, surface properties za human tissue, patient motion, endoscopic dimensions, laser safety, need for real-time measurement na validation kwa medical reference methods zinapaswa kuchunguzwa tofauti. Utafiti haujathibitisha diagnosis ya voice disorders, treatment selection au kwamba human phonation inaweza kupimwa katika clinical conditions.

Maelezo ya kina

Tatizo kuu la utafiti ni nini?

Katika contactless three-dimensional vibration measurements, vipengele vitatu vya nafasi vinavyojitegemea vya surface motion vinapaswa kubainishwa. Traditional three-dimensional laser scanning vibrometry systems zinaweza kutumia laser vibrometers tatu zinazoelekezwa kwenye vibrating surface kutoka angles tofauti. Image-based three-dimensional methods mara nyingi hutegemea high-speed cameras mbili zilizowekwa katika angles tofauti na stereoscopic triangulation.

Tatizo la pamoja la setups hizi ni kwamba vibrating surface lazima iweze kuonekana kutoka directions zaidi ya moja. Ikiwa surface iko mwisho wa narrow channel, kati ya parts nyingine au ndani ya long tube-like structure, inaweza kuwa impossible kuunda second na third viewing directions.

Synthetic vocal folds zinazotumika katika human phonation research ni mfano wazi wa tatizo hili. Vocal folds ziko ndani kabisa ya geometry ya larynx na vocal tract. Artificial vocal tracts zilizotumika katika utafiti zinaanzia 200 mm na katika baadhi ya research setups zinaweza kufikia hadi 1 metre. Katika hali hii, inaweza kuwa physically impossible kuweka stereoscopic cameras zinazouona uso kutoka angles tofauti, au camera angles zinaweza kuwa ndogo kiasi cha kupunguza depth accuracy kwa kiwango kikubwa.

Suluhisho linalopendekezwa na utafiti ni nini?

Watafiti waliunganisha principles mbili tofauti za measurement zinazofanya kazi kwenye optical axis ileile:

• Kamera ya kasi kubwa: Hupima motion components mbili ndani ya image plane.
• Single-axis laser scanning vibrometer: Hupima out-of-plane velocity component iliyo parallel na optical axis.

Kamera ya kasi kubwa hurekodi surface images. Apparent motion kati ya consecutive frames huhesabiwa katika pixel level kwa dense optical flow method. Image magnification na frame rate zikijulikana, pixel motions hubadilishwa kuwa physical velocity units.

Laser scanning vibrometer hupima Doppler shift katika backscattered component ya laser light inayotumwa kwenye surface. Doppler shift inahusiana na vibration velocity ya surface katika laser direction. Kwa kutumia galvanic mirrors, laser beam huelekezwa kwenye predefined points juu ya surface.

Figure 1 katika page 9 ya utafiti inaonyesha kwamba high-speed camera na laser vibrometer zinaangalia surface kutoka viewing axis ileile. Camera recording hufanywa kwanza, laser scanning baadaye. Measurements mbili huunganishwa kwa phase-locking kwa common periodic reference signal.

Kwa nini high-speed camera pekee haikutumika?

Single camera inaweza kupima motion katika image plane; lakini haiwezi kubaini moja kwa moja motion katika depth direction. Ingawa monocular depth estimation inaweza kufanywa kwa methods kama deep learning au perspective change, accuracy inategemea training data, geometry, surface texture na depth cues katika image.

Katika utafiti, out-of-plane motion haikukadiriwa kutoka images; ilipimwa moja kwa moja kwa laser Doppler vibrometry. Hivyo in-plane measurement, ambako camera ina nguvu, na out-of-plane measurement, ambako vibrometer ina nguvu, zilikamilishana katika mfumo mmoja.

Kwa nini laser vibrometer pekee haikutumika?

Single-axis laser vibrometer hupima velocity component katika direction ya laser beam pekee. Traditional three-dimensional vibrometry huhitaji devices tatu kutoka directions tofauti. Hii huongeza hardware cost pamoja na optical space inayohitajika kuzunguka surface.

Katika proposed method, kwa kuwa image-plane components mbili zinatolewa na camera, laser vibrometer inahitaji kupima third component pekee. Mfumo bado unahitaji measurement devices mbili, lakini zote mbili zinaangalia surface kutoka direction ileile.

Optical flow ilitumikaje?

Farnebäck dense optical flow algorithm ndani ya OpenCV ilitumika kusindika high-speed camera images. Optical flow huhesabu apparent motion ya intensity patterns katika consecutive images katika pixel level.

Kwa kuwa camera ilikuwa fixed, ilidhaniwa kwamba apparent motion iliyopimwa ilitokana na surface motion. Ili kupata reliable optical flow, surface inapaswa kuwa na high-contrast, random na trackable texture. Fine charcoal powder iliongezwa juu ya synthetic vocal folds ili kuunda random surface pattern.

Pixel displacements kutoka optical flow zilibadilishwa kuwa physical in-plane velocities kwa kutumia camera magnification calibration na frame rate.

Je, out-of-plane motion inaweza kuathiri image-plane measurement?

Object ikisogea karibu na camera au mbali nayo, image magnification inaweza kubadilika. Mabadiliko haya yanaweza kuunda apparent in-plane motion katika optical flow algorithm hata kama object haisogei horizontally au vertically.

Ili kupunguza athari hii, lens yenye focal length ya 180 mm na working distance ya takribani 500 mm ilitumika. Katika stereoscopic validation measurements, total out-of-plane motion ilionekana kuwa 5 mm au chini. Approximate relation iliyotumika katika utafiti ni hii:

\[ \varepsilon_{\mathrm{büyütme}} \approx \frac{\Delta z}{Z} \]

Hapa Δz ni out-of-plane displacement, na Z ni distance kati ya camera na surface. Kwa values zilizotumika, upper bound ya error inayotokana na magnification ilikadiriwa kuwa takribani:

\[ \frac{5\ \mathrm{mm}}{500\ \mathrm{mm}} = 0{,}01 = \%1 \]

Thamani hii si jumla ya optical-flow errors zote; ni contribution ya takribani inayotarajiwa kutokana na perspective magnification change pekee.

Camera na vibrometer grids ziliunganishwaje?

Camera data ziko kwenye image pixels, huku laser vibrometer data zikiwa kwenye scanning coordinates za device yenyewe. Ili datasets mbili ziunganishwe moja kwa moja, zinapaswa kubadilishwa kwenda common physical coordinate system.

Utafiti ulitumia reference points zenye known physical locations ambazo zinaweza kutambuliwa katika camera image na pia katika laser scanning grid. Katika example application, corners nne za camera image zilimatchiwa na outer points nne katika laser scanning grid.

Figure 2 katika page 11 ya utafiti inaonyesha processing sequence ifuatayo:

1. Kupata raw high-speed camera images,
2. Kuhesabu in-plane velocities kwa optical flow,
3. Kupima out-of-plane velocities kwenye laser scanning points,
4. Kusawazisha camera na vibrometer grids,
5. Kuinterpolate velocity components tatu kwenye common points,
6. Particle advection kupitia three-dimensional velocity vectors.

Kwa sababu ya surface motion, baadhi ya laser points zinaweza kutoka nje ya vocal fold katika nyakati fulani. Points hizi ziliondolewa kwenye dataset kwa kutumia surface mask katika camera image.

Interpolation ilifanywaje?

Kwa kuwa camera na vibrometer measurements haziko kwenye spatial points zilezile, datasets mbili ziliinterpolateiwa kwenye common virtual-particle locations. Utafiti ulitumia radial basis function, yaani RBF interpolation, pamoja na thin-plate spline kernel.

Baada ya process hii, katika kila selected surface point:

\[ \mathbf{v} = \begin{bmatrix} v_x \\ v_y \\ v_z \end{bmatrix} \]

three-dimensional vibration-velocity vector hupatikana. Locations za interpolation points zinaweza kuchaguliwa kwa uhuru ndani ya common boundaries za camera image na laser scanning region.

Tofauti kati ya Euler na Lagrange approaches ni nini?

Katika Euler approach, velocities hufafanuliwa katika fixed points angani. Surface inapopita chini ya points hizi, instantaneous velocity hupimwa katika kila point. Laser scanning vibrometry na image-based dense velocity fields mara nyingi huzalisha representation ya aina hii.

Katika Lagrange approach, virtual particle husogea pamoja na surface. Kwa kufuatilia position ya particle kwa muda, actual motion trajectory na displacement hupatikana.

Katika small vibrations, velocity field katika fixed points inaweza kutosha. Katika large shape changes kama vocal folds, ni informative zaidi kuonyesha sehemu ileile ya material ya surface imesogea wapi kwa muda.

Particle advection ilihesabiwaje?

Utafiti ulitumia stabilized trapezoidal integration scheme kuhesabu particle positions kutoka velocity field:

\[ \mathbf{d}^{\,n+1}_i-\mathbf{d}^{\,0}_i = \alpha\left(\mathbf{d}^{\,n}_i-\mathbf{d}^{\,0}_i\right) + \frac{\Delta t}{2} \left(\mathbf{v}^{\,n+1}_i-\mathbf{v}^{\,n}_i\right) \]

Hapa:

• i, ni index ya virtual particle.
• n, inaonyesha time step.
• d = [d_x, d_y, d_z]^T, ni three-dimensional displacement vector.
• v = [v_x, v_y, v_z]^T, ni three-dimensional velocity vector.
• Δt, ni time step.
• α, ni stability coefficient inayopunguza integration drift.

Watafiti walitumia α = 0,999 kwa data zao. Smaller α values zinaweza kukandamiza numerical drift kwa nguvu zaidi, lakini zinaweza kuongeza integration error katika real oscillatory components.

Figure 3 katika page 12 ya utafiti inaonyesha selection ya initial particle positions, interpolation ya velocities kwenye positions hizi, integration step na matumizi ya new particle positions katika next time step.

Kwa nini initial depth position haijulikani?

Kwa single camera, initial x na y coordinates katika image plane zinaweza kuchaguliwa. Lakini initial z coordinates za surface points kwenye camera axis haziwezi kubainishwa kutoka raw monoscopic measurement.

Kwa hiyo, ingawa monoscopic method inaweza kupima time-varying part ya out-of-plane oscillation, haiwezi kubaini peke yake absolute initial three-dimensional shape ya surface. Kwa kila particle kunaweza kuwa na depth offset isiyobadilika na muda lakini inayobadilika katika space:

\[ d_{z,\mathrm{ofset},i} = d^0_{z,i}-d^0_{z,\mathrm{gerçek},i} \]

Hii ndiyo sababu katika Figure 7 katika page 19 stereoscopic na monoscopic trajectories zinaonyesha oscillation shapes zinazofanana lakini zina tofauti katika mean depth positions. Katika comparison, mean values zilitolewa na oscillatory components pekee zikatathminiwa.

Synthetic vocal-fold setup ilifanyaje kazi?

Katika experiments, synthetic vocal folds mbili zilizotengenezwa kwa silicone rubber zilitumika. Vocal folds ziliwekwa kati ya downstream vocal tract na upstream subglottal channel. Airflow iliyotolewa kwenye system ilisababisha self-oscillation kutokana na fluid–structure interaction kati ya soft folds na air.

Figure 4 katika page 14 ya utafiti inaonyesha two-dimensional cross-section ya silencer, subglottal channel, synthetic vocal folds na vocal tract. Pia vocal-fold geometry na x, y, z coordinate system iliyotumika imeonyeshwa.

Wakati synthetic folds zina scale ya takribani 20 mm, vocal tract inaweza kuwa 200 mm au zaidi. Geometry hii hufanya iwe vigumu kuweka cameras mbili zinazoona surface kutoka angles tofauti.

Ni vifaa gani vilitumika katika monoscopic measurement?

Stereoscopic validation ilifanywaje?

Kwa validation, high-speed cameras mbili za Photron Fastcam Nova R5-4K zilizowekwa kwa angle ya takribani 40 degrees zilitumika. Cameras zilirekodi kwa 4.000 frames/second na resolution ya 2048 × 1472 pixels.

Dark markers 12 ziliongezwa kwenye upper surface ya vocal fold moja. Approximate locations za markers kwenye images zilibainishwa kwa thresholding, kisha two-dimensional Gaussian function ilifitiwa kwenye gray-level distribution ya kila marker ili kuhesabu subpixel location.

Matching markers katika cameras mbili zilitriangulateiwa katika three-dimensional space kwa kutumia calibrated camera geometry na direct linear transformation.

Figure 5 katika page 16 ya utafiti inaonyesha setups mbili tofauti zinazoangalia surface ileile:

• Setup 1: High-speed camera na laser vibrometer kwenye common optical axis.
• Setup 2: High-speed cameras mbili zenye stereoscopic angle ya takribani 40 degrees.

Vocal tract iliondolewa ili stereoscopic validation iweze kufanywa. Kwa hiyo, validation condition si sawa na narrow optical-access condition ambako mbinu inaonyesha faida yake kuu.

Kwa nini mbinu mbili hazikupimwa simultaneously?

Kwa stereoscopic marker tracking, ilipendelewa kwamba surface kuzunguka markers iwe na brightness karibu uniform. Kwa monoscopic optical flow, surface inahitaji random high-contrast texture.

Kwa kuwa imaging conditions hizi mbili zinapingana, stereoscopic measurement ilifanywa kwanza; baadaye fine charcoal powder ikaongezwa kwenye surface na monoscopic measurement ikafanywa. Kulite XCQ-093 pressure sensor inayopima subglottal pressure ilitumika kama reference katika synchronization ya experiments mbili.

Kutofanywa simultaneously kwa measurements kunaweza kusababisha vibration amplitude ya vocal folds kubadilika kidogo kati ya experiments mbili. Watafiti wanahusisha kwa sehemu monoscopic amplitudes kuwa chini kidogo katika points nyingi na reproducibility issue hii.

Operating condition ya validation experiment ni ipi?

Katika validation experiment, volumetric airflow rate ilikuwa:

\[ \dot{V}=130\ \mathrm{L\,dakika^{-1}} \]

Hali hii ilisababisha oscillation ambayo synthetic vocal folds mbili hazikugusana. Non-contact condition ilichaguliwa ili kuzuia markers kupotea kwenye mojawapo ya angled-camera images wakati wa large surface deformations.

Fundamental vocal-fold frequency katika validation condition ilikuwa takribani:

\[ f_o=140\ \mathrm{Hz} \]

.

Displacement magnitudes zilioana kwa kiasi gani?

Relative RMS difference kati ya monoscopic na stereoscopic measurements ilihesabiwa kwa equation ifuatayo:

\[ \varepsilon = \frac{d_{\mathrm{rms,m}}-d_{\mathrm{rms,s}}} {d_{\mathrm{rms,s}}} \times 100 \]

Hapa d_rms,m ni ya monoscopic method, na d_rms,s ni RMS displacement ya stereoscopic method.

Matokeo ya points 12 katika utafiti ni haya:

Licha ya pointwise differences kuwa kubwa zaidi katika out-of-plane component, average difference ya total three-dimensional displacement magnitude ilibaki %5,2. Sababu kuu ni kwamba katika points nyingi dominant motion component ilikuwa y motion katika image plane.

Phase difference ilihesabiwaje?

Time difference kati ya first cyclic peaks katika fundamental vibration frequency iligawanywa kwa vibration period:

\[ \Delta\phi = \frac{\Delta t}{T}\times100 \]

Hapa Δt ni time difference kati ya peaks katika mbinu mbili, na T ni vibration period. Result iliripotiwa si kama phase angle kwa degrees bali kama percentage ya cycle duration.

Average phase difference ya %1,1 katika in-plane components na %1,5 katika out-of-plane component inaonyesha kwamba timing ya vibration modes ilikuwa karibu katika mbinu mbili.

Time-domain graphs zinaonyesha nini?

Figure 8 katika page 20 ya utafiti inalinganisha d_x, d_y na d_z displacements katika points p1, p5, p6 na p10. Monoscopic na stereoscopic curves kwa kiasi kikubwa zina overlap kwa upande wa phase.

Amplitude differences zinaonekana zaidi katika baadhi ya points. Hasa out-of-plane d_z component katika point p5 ilipimwa kuwa ndogo zaidi kwa monoscopic method. Supplementary Figure A.13 inaonyesha comparison ileile kwa points zote 12.

Frequency-domain graphs zinaonyesha nini?

Katika Figure 9 katika page 21, frequencies zimegawanywa kwa fundamental frequency na kuonyeshwa kama f/f_o. Kwa takribani harmonics tatu za kwanza, peak locations na magnitudes za mbinu mbili kwa ujumla zinafanana.

Katika higher frequencies, noise floor ya stereoscopic measurement inakaribia harmonic amplitudes. Noise ni kubwa zaidi hasa katika out-of-plane component ya stereoscopic method. Watafiti wanahusisha hili na depth direction kuwa na uncertainty kubwa zaidi mara nyingi katika stereoscopic triangulation.

Katika monoscopic method, kwa kuwa out-of-plane component inapimwa na laser vibrometer, noise floor ya component hii ni ndogo. Kwa upande mwingine, in-plane components zinategemea camera na optical-flow processing.

Je, stereoscopic method inaweza kuchukuliwa kama ground truth?

Hapana. Utafiti unachukulia stereoscopic marker tracking kama established comparison method inayotumika kwa validation, lakini hauiwasilishi kama ground truth isiyo na error.

Uncertainties kuu katika stereoscopic system ni hizi:

• Marker brightness kutofautiana na ideal Gaussian distribution,
• Image shape ya markers kuharibika kadiri surface angle inavyobadilika,
• Camera calibration na reprojection errors,
• Higher uncertainty ya depth component katika triangulation,
• Measurements mbili kutofanywa simultaneously.

Kwa sababu vocal folds ni laini sana, kuongeza accelerometer au contact sensor nyingine kwenye surface kunaweza kubadilisha mass na mechanical properties za fold. Kwa hiyo, more accurate na non-invasive ground truth haikuweza kuundwa katika utafiti huu.

Kwa nini experiment yenye vocal tract ni muhimu?

Katika validation measurements, vocal tract iliondolewa ili kutoa nafasi kwa stereoscopic cameras. Ili kuonyesha matumizi halisi ya mbinu, vocal tract yenye urefu wa 200 mm ilirudishwa kwenye system katika experiment ya pili.

Katika condition hii, optical direction moja tu inayopita ndani ya vocal tract inaweza kutumika. Kwa hiyo stereoscopic measurement haiwezekani na proposed coaxial camera–vibrometer system inaonyesha faida yake kuu.

Conditions za experiment yenye vocal tract ni zipi?

Katika experiment hii vocal folds zinagusana. Ili kupunguza particles zilizo karibu na glottal gap inayofunguka na kufunga kusukumwa nje ya surface mask, initial particles hazikuwekwa katika range ya y = 8–10 mm. Kwa hiyo, visualized surfaces hazionekani kuungana kabisa hata kama contact ilitokea katika experiment halisi.

Whole-surface motion ilionyesha nini?

Figure 10 katika page 26 ya utafiti inaonyesha synthetic vocal-fold surface katika time instants sita katika vibration cycle moja. Surfaces zimepewa rangi kulingana na local velocity magnitude.

Katika closed phase, subglottal pressure ilijikusanya na upper surfaces zikajitokeza kuelekea direction ya airflow. Baada ya vocal folds kufunguka, pressure difference kati ya lower na upper channels ilipungua na surface ikarudi kwenye smaller out-of-plane positions.

Figure haionyeshi tu widening na narrowing ya opening, bali pia local deformation na velocity distribution katika upper surface yote. Hii inatoa surface information pana zaidi kuliko single-point vibrometer measurement.

Harmonic structure ilibadilikaje?

Figure 11 katika page 27 ya utafiti inaonyesha displacement na velocity spectra zilizowastaniwa juu ya particles zote 4.320. Fundamental frequency na higher harmonics nyingi zinaonekana wazi.

Harmonic peaks zinaweza kutambulika hadi takribani 30f_o, upper limit ya graph. Peak amplitudes zilipungua karibu linearly kwenye logarithmic graph kadiri frequency ilivyoongezeka na zikaonyesha trend inayofanana na power law.

f/f_o > 12 katika region hiyo, camera-based x na y components zina noise level inayojitenga na vibrometer-based z component. Hii inaonyesha kwamba kupima coordinate components tofauti kwa sensors tofauti pia hubadilisha noise characteristics.

Vibration modes tatu za kwanza ni zipi?

Figure 12 katika page 28 ya utafiti inaonyesha three-dimensional surface modes katika fundamental frequency, second harmonic na third harmonic:

• Mode ya kwanza, f_o: Dominant motion iko katika y direction na inawakilisha kufunguka na kufunga kwa glottal opening.
• Mode ya pili, 2f_o: Out-of-plane z motion ni dominant zaidi. Vibration node inaonekana katikati ya surface.
• Mode ya tatu, 3f_o: Out-of-plane motion bado ni dominant; regions zenye highest motion zimehamia symmetrically kutoka center kuelekea outer edges.

Kutokuwepo kwa maxima za second na third modes katikati ya surface kunaonyesha kwamba spectrum katika single measurement point inaweza isiwakilishe energy distribution katika whole vocal-fold surface.

Nguvu za utafiti ni zipi?

• Umepunguza three-dimensional full-field vibration measurement hadi single optical-access direction.
• Umetumia measurement principles mbili zinazokamilishana, camera na laser vibrometer, ndani ya clear sensor-fusion framework.
• Badala ya kukadiria pamoja in-plane na out-of-plane components kutoka image, umepima out-of-plane component moja kwa moja.
• Umeeleza processing chain kutoka raw image hadi three-dimensional surface trajectory.
• Umefafanua hatua za optical flow, grid alignment, interpolation na particle advection tofauti.
• Umeilinganisha results na method inayotegemea stereoscopic na subpixel marker tracking.
• Comparison imefanywa katika surface points 12 kwa time domain, frequency domain, RMS amplitude na phase.
• Baada ya validation, method imetumika katika vocal-tract setup yenye optical access iliyo limited kweli.
• Umeonyesha harmonic modes katika surface nzima badala ya single point.
• Umejadili uncertainties za monoscopic na stereoscopic methods tofauti.

Mipaka kuu ya utafiti ni ipi?

• Utafiti ni preprint ambayo haijapitia peer review.
• Experiments zilifanywa kwenye silicone-based synthetic models, si kwenye vocal folds halisi za binadamu.
• Hakuna human participant, patient au clinical endoscopy experiment.
• Monoscopic na stereoscopic validation measurements hazikufanywa simultaneously.
• Charcoal powder iliongezwa kwenye surface kati ya measurements na experimental condition ikajengwa upya.
• Stereoscopic method si error-free ground truth.
• Absolute initial out-of-plane position haiwezi kubainishwa kutoka monoscopic data.
• Katika Lagrange representation kunaweza kuwa na constant depth offset inayobadilika spatially.
• Validation experiment ilifanywa katika non-contact oscillation condition moja pekee na airflow ya 130 L/dakika.
• Vocal-tract application imewekewa kikomo kwa vocal tract moja ya 200 mm na condition ya 135 L/dakika.
• Synthetic tissue stiffnesses, geometries na vocal-tract resonances tofauti hazikulinganishwa systematically.
• Kwa optical flow, surface ilihitaji kuandaliwa kwa artificial contrast pattern.
• HSC na LSV measurements zilifanywa sequentially; single simultaneous setup haikuundwa.
• Kwa sababu laser vibrometer huchanganua surface point by point, transient na non-repeatable events haziwezi kupimwa katika current setup.
• Comprehensive sensitivity analysis ya RBF interpolation na measurement-grid density haikutolewa.
• Value ya α = 0,999 ilichaguliwa experimentally; generalizability yake kwa motion conditions tofauti haijajaribiwa.
• Error values zinategemea comparison katika selected marker points 12 pekee.
• Spatial error map au confidence interval kwa entire surface haikutolewa.
• Independent repeat counts na between-experiment variation hazikuripotiwa statistically.
• Data hazikuchapishwa katika open repository; ilielezwa tu kwamba zitatolewa kwa request.
• Funding na separate conflict-of-interest statement hazikutolewa katika main text.

Utafiti unaunga mkono nini?

Utafiti unaunga mkono kwamba katika periodic na repeatable vibrating surface, high-speed camera na single-axis laser scanning vibrometer data zinaweza kuunganishwa kutoka optical direction ileile ili kuunda three-component velocity field.

Utafiti pia unaonyesha kwamba kwa stabilized particle advection, velocity field hii inaweza kubadilishwa kuwa oscillatory three-dimensional displacement trajectories.

Katika synthetic vocal-fold validation, reconstruction ya total displacement magnitude kwa average difference ya %5,2 na small phase differences inaunga mkono kwamba method kwa ujumla ilihifadhi vibration amplitude na mode timing katika condition iliyochunguzwa.

Experiment yenye vocal tract inaonyesha kwamba three-dimensional vibration modes za entire surface zinaweza kupatikana katika geometry ambako different viewing angles hazipatikani.

Utafiti haujathibitisha nini?

Utafiti haujathibitisha kwamba method itafanya kazi kwa accuracy ya %5,2 katika vibrating surfaces zote. Value hii ni average difference ya total three-dimensional displacement magnitude katika points 12; out-of-plane difference ilizidi %20 katika baadhi ya points.

Utafiti hauonyeshi kwamba single viewing angle ina maana single device, low cost au easy setup. High-speed camera, high-precision laser vibrometer, synchronized reference sensor na extensive data processing zinahitajika.

Haijaonyeshwa kwamba method inaweza kupima transient, non-periodic au non-repeatable vibrations kati ya experiments kwa current sequential scanning setup.

Utafiti hauvalidate three-dimensional measurement ya vocal folds halisi za binadamu, patients au speech disorders.

Haijaonyeshwa kwamba measured vibration modes zinahusiana na voice disorder fulani, voice quality au treatment outcome.

Utafiti haujathibitisha kwamba monoscopic method kwa ujumla ni accurate zaidi kuliko stereoscopic method. Mbinu mbili zina error sources tofauti na stereoscopic comparison pia si ground truth.

Umuhimu wake kwa yaliyopita, sasa na yajayo ni upi?

Kwa upande wa yaliyopita: Three-dimensional vibration measurements mara nyingi zilifanywa kwa multiple cameras au laser vibrometers tatu katika angles tofauti. Single-camera solutions zilihitaji depth estimation, assumptions specific kwa geometry fulani au learning-based models.

Mchango wa sasa: Utafiti unaunganisha established measurement principles mbili kwenye common axis, ukipima depth component moja kwa moja kwa laser na other components mbili kwa image processing. Hivyo unapunguza optical-access requirement hadi direction moja badala ya kupunguza hardware hadi sensor moja.

Kwa upande wa yajayo: Watafiti wanapendekeza kubadilisha high-speed camera na vibrometer kuwa single simultaneous setup kwa kutumia semi-transparent mirror. Continuous-scanning au multi-beam vibrometers zinaweza kupunguza measurement time. Hardware ikipunguzwa ukubwa wa kutosha na kuwa safe, research kuelekea endoscopic experiments inaweza kufanywa. Hata hivyo, clinical na live-human application si direct result ya study hii.

Mbinu na Matokeo ya Utafiti

Muhtasari wa kiufundi wa research design

Separate experimental conditions

Main validation results

Main findings za vocal-tract application

Taarifa ya kisayansi inayobebwa na figures

• Page 9, Figure 1: Inaonyesha sequential measurement setup ya high-speed camera na laser vibrometer kwenye optical axis ileile.
• Page 11, Figure 2: Inaonyesha processing chain kutoka raw image kwenda optical flow, LSV grid, alignment na particle advection.
• Page 12, Figure 3: Inaeleza interpolation ya velocities kwenye particle positions na integration ya positions kwa muda.
• Page 14, Figure 4: Inaonyesha synthetic larynx setup, vocal-tract location na vocal-fold geometry.
• Page 16, Figure 5: Inalinganisha monoscopic coaxial setup na two-camera stereoscopic validation setup.
• Page 17, Figure 6: Inaonyesha locations za validation markers kumi na mbili kwenye vocal fold.
• Page 19, Figure 7: Inaonyesha three-dimensional trajectories za mbinu mbili zinafanana kwa shape lakini zina offset katika mean z position.
• Page 20, Figure 8: Inaonyesha time-domain amplitude na phase comparison katika selected points nne.
• Page 21, Figure 9: Inaonyesha spectral agreement katika fundamental frequency na first harmonics pamoja na noise differences katika high frequencies.
• Page 26, Figure 10: Inaonyesha opening, closing na velocity distribution ya entire vocal-fold surface katika cycle moja huku vocal tract ikiwa imewekwa.
• Page 27, Figure 11: Inaonyesha displacement na velocity spectra zilizowastaniwa juu ya whole surface.
• Page 28, Figure 12: f_o, 2f_o na 3f_o Inaonyesha three-dimensional surface structures za vibration modes.
• Pages 39–40, Supplementary Figures A.13–A.14: Zinatoa time- na frequency-domain comparisons kwa points zote kumi na mbili.

Taarifa zinazokosekana kwa upande wa reproducibility

Utafiti unaeleza devices, sampling rates, image resolutions, number of measurement points, optical-flow algorithm, interpolation kernel na integration coefficient. Hata hivyo, taarifa zifuatazo hazijaripotiwa kikamilifu:

• Ni independent experimental repeats ngapi zilifanywa katika kila condition,
• Between-experiment standard deviation au confidence interval,
• Exact OpenCV settings za optical-flow parameters,
• RBF scale na regularization parameters,
• Physical coordinates zote za laser scanning grid,
• Open access kwa raw image na vibrometer data,
• Open-repository link ya entire processing code.

Maelezo ya Chanzo na Mbinu

Jina kamili la asili la utafiti: A Monoscopic, Three-Dimensional Vibration Measurement Method and its Application to Human Phonation Research

Waandishi, kwa mpangilio sahihi: Christoph Heigl; Anja Böttger; Stefan Kniesburges; Stefan Becker.

Co-first author au equal contribution: Hakuna statement ya co-first authorship au equal contribution katika text.

Corresponding author: Christoph Heigl — christoph.heigl@fau.de.

Institutional affiliations:

• Christoph Heigl, Anja Böttger na Stefan Becker: Institute of Fluid Mechanics, Friedrich-Alexander-Universität Erlangen-Nürnberg, Cauerstraße 4, 91058 Erlangen, Bavaria, Ujerumani.
• Stefan Kniesburges: Division of Phoniatrics and Pediatric Audiology, Department of Otorhinolaryngology, Head & Neck Surgery, University Hospital Erlangen, Medical School, Friedrich-Alexander-Universität Erlangen-Nürnberg, Waldstraße 1, 91054 Erlangen, Bavaria, Ujerumani.

DOI:10.2139/ssrn.7201217

SSRN record number: 7201217.

Jarida: Peer-reviewed journal name au accepted final journal version haijathibitishwa.

Original journal publisher: Hakuna verified peer-reviewed journal publisher kwa version hii.

Publication platform: SSRN.

Publication year: 2026.

ResearchGate full-text upload date: 30 Julai 2026.

Idadi ya kurasa: File ya pages 40; inajumuisha highlights, main study, references na supplementary comparison graphs.

Aina ya chanzo: Original experimental measurement-method development, validation na application research; preprint.

Hali ya peer review: Pages za utafiti zina statements “This preprint research paper has not been peer reviewed” na “Preprint not peer reviewed”. Utafiti haupaswi kuwasilishwa kama peer-reviewed journal article.

Official study link:SSRN study record

Upatikanaji wa data: Waandishi wameeleza kwamba data zitatolewa kwa request. Hakuna open-data repository au raw-data download link iliyotolewa.

Ufadhili: Main text haijaripoti funding organization au project number.

Mgongano wa maslahi: Main text haina separate conflict-of-interest statement. Hii haimaanishi kwamba conflict of interest haipo kwa uhakika; inaonyesha tu kwamba declaration haipo katika text.

Hali ya kimaadili: Hakuna human au animal participant aliyetumika katika utafiti. Experiments zilifanywa kwenye silicone-based synthetic vocal-fold models. Separate ethics-board approval haijaripotiwa.

Michango ya waandishi:

• Christoph Heigl: Conceptualization, methodology, software, experiment, visualization, initial draft, review na editing.
• Anja Böttger: Experiment, software, review na editing.
• Stefan Kniesburges: Conceptualization, initial draft, review na editing.
• Stefan Becker: Supervision, resources, review na editing.

Maelezo haya ya Kiswahili yameandaliwa kwa kutegemea main text, equations, experimental setups, method diagrams, three-dimensional trajectory graphs, time- na frequency-domain comparisons, vocal-fold surface visualizations, vibration-mode analyses, supplementary figures na results za study. Clinical diagnosis, treatment, human-health au commercial-device-performance results ambazo hazipo katika study hazijaongezwa.

External sources zilitumika tu kwa bibliographic verification ya DOI, SSRN record, upload date na institutional affiliations. Scientific method, numerical results na limitations zinategemea study yenyewe.

Interpretation limit muhimu zaidi ya study ni kwamba phrase “single viewing angle” haimaanishi single sensor. Method inategemea data fusion ya high-speed camera na laser scanning vibrometer kwenye common optical axis.

Limit ya pili muhimu ni kwamba average difference ya %5,2 si single accuracy value inayotumika kwa coordinates zote na points zote. Value hii ni average difference ya three-dimensional displacement magnitude katika points 12. Average difference ya out-of-plane component ni %10,0 na largest pointwise difference ni %21,7.

Limit ya tatu muhimu ni kwamba absolute initial depth haiwezi kupimwa. Method inaweza kubaini oscillatory out-of-plane motion, lakini haiwezi kutoa full three-dimensional shape ya initial surface bila additional geometry measurement.

Utafiti huu haujathibitisha kwamba method inaweza kutumika kwa real human phonation, clinical endoscopy au patient assessment. Applications hizi zinahitaji independent validation kuhusu living tissue, safety, motion artifact, surface texture, device miniaturization na clinical reference methods.