Assessing garment stretch through low-cost remote color reflectance sensing for smart textile applications

Abstract

In this study, we explore a new method for quantifying stretch/strain of fabrics by remotely measuring the color properties of the fabric under static, stepwise strain conditions as it stretches/relaxes. This method is based on a colloquially well-understood phenomena—that as an elastic fabric (e.g., a knit) stretches, the reflected color content changes, revealing more of the underlying surface. Here, we demonstrate that for a variety of elastic knit fabrics of varying color (white, black, red), when stretched in front of varying color backgrounds (white, black, and red), the reflected color content (measured via both RGB Euclidean distance and ΔE, captured with an off-the-shelf camera) correlates to the percentage stretch of the fabric. This sensitivity varies depending on the specific fabric-background color pairing and generally decreases beyond approximately 30% strain as the ΔE response begins to plateau. This method offers a new strategy for remotely, and non-invasively, measuring strain in elastic fabrics without the need for additional embedded sensors/hardware.

1 Introduction

The development of reliable, low-cost methods for monitoring how garments behave during motion /deformation/wear is becoming increasingly important in wearable technology applications, such as soft-robotic exoskeletons and dynamic compression garments (Fu et al., 2022; Sanchez et al., 2021; Xiong et al., 2021). In all these systems, garments are not merely passive layers but rather actively interact with the body by applying controlled forces or sensing any biomechanical changes. A major challenge in these applications is the measurement of fabric deformation, particularly stretch or strain, which is essential for informing control systems and ensuring that mechanical forces are accurately distributed to the body in therapeutic garments applications (Granberry et al., 2021; Granberry et al., 2019).

Despite the importance of obtaining accurate and reliable strain measurement in smart textile systems, particularly for wearable sensing and actuation applications, capturing strain in textiles remains non-trivial due to the non-linear and heterogeneous nature of fabric structures, especially in knitted materials, where stretching involves complex loop reconfiguration rather than simple material elongation. Currently, two primary strategies exist for measuring fabric strain: embedded strain sensors and digital image correlation (DIC)/motion capture systems. Embedded sensors-such as conductive elastomers or strain-sensitive stitch sensors are integrated into the fabric to directly measure deformation through electrical property changes (Gioberto and Dunne, 2012; Seyedin et al., 2019; Wang et al., 2020). Although effective, these approaches require additional materials and fabrication steps and require direct integration of these materials into the textile/wearable system. These additional fabrication steps and materials can alter the mechanical behavior of the textile and limit their adaptability across different garment types and use scenarios.

On the other hand, motion capture systems, such as those utilizing infrared camera arrays, and digital image correlation (DIC) systems provide remote strain estimation by tracking surface features. Systems such as the infrared camera arrays (e.g., Vicon systems) or stereo imaging setups (e.g., VIC-3D) rely on markers or speckle patterns applied to the fabric surface to compute displacement and strain fields (de Aguiar et al., 2007; Dunne et al., 2011; Lennard and Dulieu-Barton, 2014; Yang et al., 2018). Although these approaches offer high accuracy and full-field strain capabilities, they are typically confined to laboratory settings and involve costly hardware and cumbersome setup, limiting their usability in real world environments (Shaheen et al., 2009; Vlasic et al., 2007).

Beyond these established methods, the concept of inferring mechanical deformation through optical property changes has been explored in other domains. For instance, variations in light reflectance have been used to estimate skin stretch where deformation alters surface geometry and optical scattering characteristics. Guzelsu et al. (2003) demonstrated that reflectivity of polarized light intensity increases linearly with skin stretch in both in vitro pig skin and in vivo human skin, with changes resulting from alterations in surface roughness across skin layers (Guzelsu et al., 2003). Federici et al. (1999) also showed that reflectivity increases with stretch due to decreased interface roughness producing smoother reflective surfaces (Federici et al., 1999). More recently, Li et al. (2024) found that optical properties closely mirrored mechanical property changes during prolonged skin stretching (Li et al., 2024). With the underlying principle that mechanical deformation can include measurable changes in optical response, this suggests an opportunity to extend a similar concept to the textile system.

In knitted textiles in particular, deformation is accompanied not only by geometric elongation but also by changes in surface topology, yarn spacing, and light interaction. As knitted fabrics stretch, the opening of interlooped yarn gaps can modify how light is being reflected and absorbed through the material which often shows as a visible change in color intensity or saturation. Also, this is a phenomenon with which many of us have practical experience: when wearing tight-fitting elastic clothing, depending on the color of the fabric. Here, we seek to leverage this exact phenomenon—the fact that elastic fabric color quality often changes as it stretches—to develop a new remote measurement strategy of garment strain using simple remote cameras/algorithms to enable non-contact strain monitoring in wearable applications.

In this study, we propose a novel, low-cost, alternative for remote measurement of textile strain using a remote camera to infer fabric stretch by analyzing changes in color content of the garment surface as it stretches and relaxes. Specifically, we hypothesize that as knit fabrics stretch, their surface structure alters the visible reflection of light, modifying how colors appear. By remotely recording these color changes with a camera and analyzing the changing color values, we demonstrate that it is possible to estimate fabric strain without embedding sensors or placing any physical markers on the fabric.

2 Methods

2.1 Materials and equipment

To investigate the relationship between fabric stretch and perceived color change, we used a combination of mechanical testing equipment, optical imaging tools, and controlled lighting. Fabric samples were stretched using an Instron 3,365 Tensile Tester with a 100 N load cell and pneumatic clamps. Illumination was provided by a 12″ × 12″ Glendan PBST-100 light box with 5,500 K, >95 (CRI) color rendering index. To accommodate the tensile tester, a hole was cut into the bottom of the light box so it could fit over the bottom clamp; the light box already had a hole at the top.

We used three PVC backdrops in different colors: white, black, and red to investigate background effects on color readings. Backdrop colors were measured using the same imaging and averaging procedure as the fabric samples (Section 2.4), with no sample present. Under the experimental lighting and camera settings, the backdrops measured RGB (173, 172, 172), (151, 148, 149), (213, 5, 10) for the white, black, and red backdrops, respectively. The elevated values for the black backdrop reflect the light source being positioned between the sample and the backdrop, which brightened its appearance under imaging conditions. Samples for the imaging experiments were made from a single-jersey knit fabric (95% polyester, 5% spandex) in three matching colors (white, black, and red). Each imaging sample was cut to 165 mm × 75 mm and stretched in the course (most extensible) direction. This size was selected so the fabric fully filled the camera frame with margin for the lens diffuser cap, allowing images to be captured without visible edges or shadows.

The fabric’s material properties were characterized using separate specimens cut from the same yardage. Fabric weight was measured as directed in ASTM D3776 (Option C) using a single 10 × 10 cm specimen, excluding selvages, yielding 158 g/m². Thickness was measured as directed in ASTM D1777 using a compressometer, yielding 0.3 mm. Stretch percentage was measured as directed in ASTM Test Method D2594 using an Instron tensile testing machine (model number 3365) using specimens of the dimensions specified in that standard, yielding 242%.

Images were captured using an iPhone SE (2nd generation), 12MP wide-angle camera with ƒ/1.8 aperture. A Moment M-Series 10X macro lens with diffuser hood was attached to the phone to maintain consistent imaging conditions and minimize glare or shadows during photography.

2.2 Test conditions

Each test condition involved a combination of fabric and backdrop color. For each fabric color, three physically separate samples were prepared, and each sample was tested against each of the three backdrops (white, black, red), resulting in 27 test conditions. Two trials were conducted per sample per backdrop, yielding six measurements (three samples × two trials) per fabric-backdrop condition. Each sample was stretched from 0 to 100% strain and then returned to 0% with photos captured at every 10% strain increment. This produced a total of 21 photos per test condition (11 during stretching, 10 during relaxation). Figure 1 shows the stretching sequence (0 to 100%) for sample 1, trial 1 for each fabric and backdrop combination.

Figure 1

2.3 Experimental procedure

The experimental setup involved mounting the light box over the tensile tester so that the clamps could pass through the top and bottom holes. A PVC backdrop was placed vertically at the rear of the light box. Fabric samples were clamped in the course direction with a fixed gage length of 100 mm between clamps; 10 mm of clamp displacement therefore corresponded to 10% strain. The iPhone, fitted with the macro lens and the diffuser hood, was mounted on a tripod inside the light box, with the lens horizontally centered over the sample at 0% strain and positioned at a fixed distance of 1 inch from the sample.

All photographs were captured using the camera module in the Adobe Lightroom app and exported as JPEG images. Exposure compensation (0.0) and white balance were locked after an initial white-backdrop calibration; the aspect ratio was fixed at 1:1. ISO was left on automatic and therefore varied with scene luminance across conditions, ranging from ISO 20 for the white-fabric conditions to ISO 80-320 for the black-fabric conditions, with red-fabric conditions falling between ISO 64-100. Within a given stretch cycle, ISO generally decreased from its initial value as backdrop show-through increased scene brightness. Illumination (5,500 K, CRI > 95) was otherwise held constant throughout.

At the start of each test, a baseline photo was taken with the fabric at 0% strain. Clamp displacement was then increased in 10 mm increments, corresponding to 10% strain intervals. At each increment the fabric was held until the measured force stabilized (typically within a few seconds) before the photograph was taken. Once the fabric reached 100% strain, the process was reversed in 10 mm decrements, with a photograph taken at each step, and a final image captured at 0% strain to complete the stretch-relaxation cycle. The fabric did not fully return to its initial resting appearance at the moment the final 0% photograph was taken; this is reflected in the elevated ΔE values at the final 0% point across all conditions in the hysteresis data (Figure 2). All testing remained within the fabric’s elastic range, as samples were strained to a maximum of 100%, well below the measured stretch capacity of 242%.

Figure 2

2.4 Image post processing

For each fabric-backdrop combination, the mean and standard deviation of the Euclidean distance and ΔE values were calculated at each strain level across all samples and trials to quantitatively compare how fabric color appearance shifted with strain and how different backdrops influenced this change.

3 Results

3.1 Visual and quantitative color change with strain

Figure 1 shows, for each fabric-backdrop combination, the photographed fabric at each strain increment together with its averaged pixel color and the corresponding ΔE relative to the unstrained (0%) state. As strain increased, the composite color visibility shifted, blending the fabric and backdrop colors as the knit opened and revealed more of the underlying surface. This visual change and increase in ΔE was concentrated in the lower strain range, with the majority of the color change occurring below 30% strain. Beyond 30% strain, the response largely plateaued and the swatches changed little with further stretching. The magnitude of change depended strongly on the fabric-backdrop pairing, being most pronounced for high-contrast combinations (e.g., black fabric on a red backdrop) and minimal for color-matched combinations (e.g., white fabric on a white backdrop).

3.2 ΔE vs. fabric strain and hysteresis

Both RGB Euclidean Distance and ΔE values were calculated for each fabric-backdrop-strain condition. Since both parameters exhibited similar correlation with fabric strain, we simplified the results and analysis by focusing solely on the ΔE vs. strain relationship. Figure 2 presents mean ΔE versus strain (0–100, 10% increments) for each fabric-backdrop combination, with shaded ±1 SD bands for the loading curve. Across all fabric-backdrop combinations, ΔE increased with strain, confirming that fabric deformation produced measurable changes in optical appearance. However, the sensitivity of the ΔE response varied depending on the specific fabric and backdrop color combination. Consistent with Figure 1, the ΔE-strain response rose steeply at low strain and largely plateaued beyond approximately 30% strain.

Figure 2 also shows the cyclic response. Stretch and relaxation ∆E values were closely matched across most of the strain range, indicating minimal hysteresis during a single loading-unloading cycle. The exception was at the final 0% point, where ∆E remained elevated across all conditions, reflecting that the fabric had not fully returned to its initial resting appearance at the moment the final image was captured. Assessing durability under prolonged and repeated cyclic deformation remains an area for future work.

3.3 Statistical analysis

3.3.1 Effect of fabric and backdrop color on ΔE

To determine whether fabric color, backdrop color, and their interaction significantly influenced the optical response, two-way ANOVAs were performed at each strain level (10–100%). Significant main effects of fabric color and backdrop color were observed across all strain levels, along with significant fabric color × backdrop interactions (Table 1). Partial eta-squared values indicated consistently large effect sizes for all three effects (ηp² = 0.652–0.950). Fabric and backdrop color each reached their largest effect at 40% strain (fabric: F (2, 45) = 430.73, p = 4.539 × 10⁻³, ηp² = 0.950; backdrop: F (2, 45) = 430.76, p = 4.531 × 10⁻³, ηp² = 0.950) and the interaction was largest at 80% strain (F (4, 45) = 193.76, p = 9.625 × 10⁻²⁸, ηp² = 0.945). These results show that the optical response is strongly dependent on the combined fabric-backdrop color relationship rather than just fabric or backdrop color alone.

3.3.2 Accuracy of ΔE-based strain estimation

While the ANOVA results demonstrate that ΔE is significantly affected by strain, fabric color, and backdrop color, statistical significance alone does not indicate how accurately ΔE can estimate strain. We therefore performed regression analyses to quantify predictive accuracy, using all six observations per strain level and assessing model performance by the coefficient of determination (R²), mean absolute error (MAE), and root mean square error (RMSE) over both the full strain range (0–100%) and the restricted low-strain range (0–30%) (Table 2).

Prediction accuracy varied across fabric–backdrop combinations. Over the full 0–100% strain range, R² values ranged from 0.13 to 0.86, with MAE ranging from 9% to 24% strain and RMSE ranging from 11% to 29% strain; the best performance was for white fabric on a red backdrop and the worst for black fabric on a black backdrop. Restricting the analysis to 0–30% strain improved accuracy substantially for most combinations: seven of the nine combinations achieved R² > 0.80, with the best reaching R² = 0.95 and MAE and RMSE as low as 2% strain. This improvement reflects the greater optical sensitivity at low strain and the saturation of the ΔE response beyond ~30% strain, which makes the color–strain relationship non-invertible at higher strains.

4 Discussion

In Figure 2, we see clear relationships between strain percentage and ΔE values for most fabric color/backdrop color combinations. In conditions with either the white backdrop or red backdrop, we see a clear difference in ΔE vs. strain behavior depending on the fabric color. In the white backdrop condition we see distinct ΔE sensitivity for both the black and red fabrics, most notably at small strains (≤30% strain), and nearly zero sensitivity for the white fabric (which is to be expected when the fabric and backdrop colors match). In the red backdrop condition we see sensitivity for all three fabric colors, again primarily limited to small strains (≤30% strain). In the black backdrop condition we see the least ΔE sensitivity for all fabric colors, though the red fabric continued to increase at higher strains, up to 100% strain, albeit with greater variability than the other conditions (with nearly no sensitivity detected in the white and black fabrics). These observations demonstrate that the effectiveness of the proposed optical strain estimation method depends strongly on the optical contrast between fabric and its background.

Among all imaging conditions, the red backdrop, surprisingly, produced the most striking sensitivity results of all conditions (even for the red fabric/red backdrop combination). To better understand this result, we can see in Figure 1 the visually perceptual differences in image color as all three fabric colors are stretched in front of the red background, reflected in the averaged pixel colors shown below each photograph. It is clear from both these images and average pixel representations that even at small strains, the composite color evolves as a mixture of the foreground fabric color and the red background color, and increasing strain results in a progressive color mixing effect in the resulting image/pixel. This confirms our initial hypothesis (and the colloquial/common knowledge) that fabric stretch changes the resulting fabric color properties, and this relationship is mediated both by fabric color and the underlying backdrop color.

Across all tested fabric–background combinations, the ΔE-strain relationship exhibited a gradual plateau beginning at approximately 30% strain. As the response approached this plateau, additional stretching produced smaller incremental changes in ΔE, reducing the sensitivity of the optical measurement at higher strain levels. Consequently, while ∆E continued to change modestly beyond 30% strain, prediction accuracy and sensitivity were greatest within the lower strain range and became increasingly dependent on the specific fabric–background color combination at higher strains. This behavior is likely related to the knitted fabric structure approaching a configuration in which further stretching results in relatively small changes in the visible fabric–background pattern.

While the ANOVA results demonstrate that ΔE is significantly influenced by strain, fabric color, backdrop color, and their interaction, the regression analysis provides a more practical assessment of the proposed method by quantifying the prediction performance. Consistent with the contrast-dependance described above, high-contrast fabric-backdrop combinations consistently produced higher R² values and lower prediction errors than low-contrast combinations. This trend indicates that sufficient color contrast is necessary to obtain reliable ΔE measurements for quantitative strain estimation. In addition, prediction accuracy improved substantially within the lower range (0–30%), corresponding to the region where the ΔE-strain relationship exhibited the greatest sensitivity.

Interestingly, the hysteresis data presented in Figure 2 shows a slight difference in ΔE values between stretch and relaxation of the textile within the tested stretch-relaxation cycle evaluated in this study. This observation suggests that the color response is relatively consistent during loading and unloading across most of the strain range under evaluated conditions. Conceptually, this behavior may be attributed to the fact that the composite color effect depends primarily on the relative negative space between the textile elements (which should be similar for both stretch and relaxation directions for a given strain percentage). This is notably different from many other mechanical/electrical textiles parameters (e.g., stress, resistance, etc.), which often demonstrate considerable hysteresis during strain and subsequent relaxation (Abbott and Grosberg, 1966; Liang et al., 2019; Matsuo and Yamada, 2009). However, additional investigation involving repeated and prolonged cyclic loading is needed to determine whether ΔE-strain relationship remains stable over prolonged deformation and repeated stretch-relaxation cycles.

An important outcome of this study is the establishment of quantitative performance guidelines for optical strain estimation across different fabric–background combinations. The reported R², MAE, and RMSE values provide objective measures of prediction quality and can be used to evaluate the optical contrast between fabric and its backdrop before strain estimation. If insufficient contrast is detected, the system could alert the user or recommend an alternative imaging configuration. Although this study was conducted under controlled laboratory conditions, it establishes the feasibility of estimating textile strain using optical color changes without embedded sensors or physical surface markers. These findings provide practical guidance for selecting imaging conditions and calibration models for wearable textile strain sensing applications.

5 Limitations and future work

This proof-of-concept work included considerable experimental control of environmental parameters (e.g., camera angle relative to the fabric was fixed, macro lenses were deployed to magnify the fabric frame, lighting conditions were precisely controlled, etc.), which likely represented best-case scenario test conditions. If similar tests were conducted under less ideal conditions, it is unknown whether (and to what extent) the results would differ. In addition, ISO was not manually fixed during image capture and varied with the luminance of each fabric–backdrop condition; because images were saved as in-camera-processed JPEGs, this variation may have introduced minor condition-dependent differences in rendered color. Locking ISO and capturing raw images would remove this source of variability. We also only tested 3 different fabric colors (of the same fabric type), and 3 different backdrop colors; additional testing would be required to better understand the relationship between different fabric types, fabric colors, backdrop colors, and lighting conditions. The method also depends on sufficient optical contrast between fabric and backdrop. Low-contrast combinations produced weak ∆E responses and correspondingly poor strain-estimation accuracy, and accuracy was highest within the lower strain range (≤30%), where the optical response was most sensitive. Finally, testing in this study only focused on uniaxial strain, with the camera oriented perpendicularly to the strain axis. Less optimal camera angles or more complex strains (e.g., 2D strain and/or out-of-plane strain) require additional study to understand the impacts of those test parameters on system performance/sensitivity.

Several directions would extend this work beyond the present laboratory condition. Comparison with digital image correlation (DIC) and other marker-based strain-measurement techniques would provide an important benchmark for evaluating the proposed method against established full-field strain methods. Future implementations will also focus on developing adaptive calibration procedures and image-processing algorithms that improve robustness under varying background and lighting conditions. Finally, extending the approach to more complex garment deformation modes, including biaxial and out-of-plane stretching, will be important for real-world use. Biaxial tensile experiments would quantify the optical response under multi-directional strain, and curved substrate testing would evaluate performance under non-planar deformation conditions that more closely represent garment wear.

6 Conclusion

This study demonstrated the feasibility of a low-cost, remote, and non-invasive method for estimating fabric strain using ΔE measurements captured using a standard iPhone camera. The results quantitatively confirm the commonly observed phenomenon that elastic fabrics produce visually perceptible color changes as they stretch, and that the relative degree of color change depends on both the fabric color/properties and the underlying backdrop color properties. Furthermore, the results show that the predictive performance of the proposed method is strongly influenced by the fabric-backdrop color combination, with the highest accuracy under high contrast imaging conditions and at lower strain levels where the ΔE response is most sensitive. The proposed approach provides a foundation for future development of non-contact strain monitoring methods for textile-based sensing and actuation applications (e.g., soft exoskeletons, dynamic compression garments, wearable stretch sensor garments, etc.).

References

• 1

  AbbottG. M.GrosbergP. (1966). Measurement of fabric stiffness and hysteresis in bending. Text. Res. J.36, 928–930. doi: 10.1177/004051756603601012

  • CrossRef
  • Google Scholar

• 2

  AguiarE.deTheobaltC.StollC.SeidelH.-P. (2007) Marker-less Deformable Mesh Tracking for Human Shape and Motion Capture., in 2007 IEEE Conference on Computer Vision and Pattern Recognition, (New York, NY, USA: IEEE), 1–8. doi: 10.1109/CVPR.2007.383296

  • CrossRef
  • Google Scholar

• 3

  DunneL. E.GiobertoG.KooH. (2011) A Method of Measuring Garment Movement for Wearable Sensing., in 2011 15th Annual International Symposium on Wearable Computers, (New York, NY, USA: IEEE), 11–14. doi: 10.1109/ISWC.2011.19

  • CrossRef
  • Google Scholar

• 4

  FedericiJ. F.GuzelsuN.LimH. C.JannuzziG.FindleyT.ChaudhryH. R.et al. (1999). Noninvasive light-reflection technique for measuring soft-tissue stretch. Appl. Opt.38, 6653–6660. doi: 10.1364/AO.38.006653,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 5

  FuC.XiaZ.HurrenC.NilghazA.WangX. (2022). Textiles in soft robots: current progress and future trends. Biosens. Bioelectron.196:113690. doi: 10.1016/j.bios.2021.113690,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6

  GiobertoG.DunneL. (2012) Theory and characterization of a top-thread coverstitched stretch sensor., in 2012 IEEE International Conference on Systems, Man, and Cybernetics (SMC), (New York, NY, USA: IEEE), 3275–3280. doi: 10.1109/ICSMC.2012.6378296

  • CrossRef
  • Google Scholar

• 7

  GranberryR.ComptonC.WoelfleH.BarryJ.HolschuhB. (2021). Enhancing performance and reducing wearing variability for wearable technology system–body interfaces using shape memory materials. Flex. Print. Electron.6:024004. doi: 10.1088/2058-8585/abf848

  • CrossRef
  • Google Scholar

• 8

  GranberryR.EschenK.HolschuhB.AbelJ. (2019). Functionally graded knitted actuators with NiTi-based shape memory alloys for topographically self-fitting wearables. Adv. Mater. Technol.4:1900548. doi: 10.1002/admt.201900548,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9

  GuzelsuN.FedericiJ. F.LimH. C.ChauhdryH. R.RitterA. B.FindleyT. (2003). Measurement of skin stretch via light reflection. J. Biomed. Opt.8, 80–86. doi: 10.1117/1.1527936,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 10

  LennardF.Dulieu-BartonJ. M. (2014). Quantifying and visualizing change: strain monitoring of tapestries with digital image correlation. Stud. Conserv.59, 241–255. doi: 10.1179/2047058413Y.0000000097

  • CrossRef
  • Google Scholar

• 11

  LiK.-Y.LineC.-C.ChenC.-C.ChenS.-Y. (2024). Reveal mechanical and optical properties changes during prolonged skin stretching via in vivo continuous investigation. J. Appl. Phys.136:144703. doi: 10.1063/5.0221718

  • CrossRef
  • Google Scholar

• 12

  LiangA.StewartR.Bryan-KinnsN. (2019). Analysis of sensitivity, linearity, hysteresis, responsiveness, and fatigue of textile knit stretch sensors. Sensors19:3618. doi: 10.3390/s19163618

  • CrossRef
  • Google Scholar

• 13

  MatsuoM.YamadaT. (2009). Hysteresis of tensile load—strain route of knitted fabrics under extension and recovery processes estimated by strain history. Text. Res. J.79, 275–284. doi: 10.1177/0040517507090504

  • CrossRef
  • Google Scholar

• 14

  SanchezV.WalshC. J.WoodR. J. (2021). Textile technology for soft robotic and autonomous garments. Adv. Funct. Mater.31:2008278. doi: 10.1002/adfm.202008278

  • CrossRef
  • Google Scholar

• 15

  SeyedinS.ZhangP.NaebeM.QinS.ChenJ.WangX.et al. (2019). Textile strain sensors: a review of the fabrication technologies, performance evaluation and applications. Mater. Horiz.6, 219–249. doi: 10.1039/C8MH01062E

  • CrossRef
  • Google Scholar

• 16

  ShaheenM.GallJ.StrzodkaR.Van GoolL.SeidelH.-P. (2009) A comparison of 3d model-based tracking approaches for human motion capture in uncontrolled environments., in 2009 Workshop on Applications of Computer Vision (WACV), (New York, NY, USA: IEEE), 1–8. doi: 10.1109/WACV.2009.5403039

  • CrossRef
  • Google Scholar

• 17

  VlasicD.AdelsbergerR.VannucciG.BarnwellJ.GrossM.MatusikW.et al. (2007). Practical motion capture in everyday surroundings. ACM Trans. Graph.26:35. doi: 10.1145/1276377.1276421

  • CrossRef
  • Google Scholar

• 18

  WangJ.LuC.ZhangK. (2020). Textile-based strain sensor for human motion detection. Energy Environ. Mater.3, 80–100. doi: 10.1002/eem2.12041

  • CrossRef
  • Google Scholar

• 19

  XiongJ.ChenJ.LeeP. S. (2021). Functional fibers and fabrics for soft robotics, wearables, and human–robot interface. Adv. Mater.33:2002640. doi: 10.1002/adma.202002640,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 20

  YangJ.FrancoJ.-S.Hetroy-WheelerF.WuhrerS.. (2018) Analyzing Clothing Layer Deformation Statistics of 3D Human Motions., in Computer Vision – ECCV 2018, eds. V. Ferrari, M. Hebert, C. Sminchisescu, and Y. Weiss (Cham: Springer International Publishing), 245–261. doi: 10.1007/978-3-030-01234-2_15

  • CrossRef
  • Google Scholar