Search Results

You are looking at 1 - 10 of 61 items for :

  • "signal processing" x
  • Refine by Access: All Content x
Clear All
Restricted access

Mickaël Begon and Patrick Lacouture

A marker has to be seen by at least two cameras for its three-dimensional (3-D) reconstruction, and the accuracy can be improved with more cameras. However, a change in the set of cameras used in the reconstruction can alter the kinematics. The purpose of this study was to quantify the harmful effect of occlusions on two-dimensional (2-D) images and to make recommendations about the signal processing. A reference kinematics data set was collected for a three degree-of-freedom linkage with three cameras of a commercial motion analysis system without any occlusion on the 2-D images. In the 2-D images, some occlusions were artificially created based on trials of real cyclic motions. An interpolation of 2-D trajectories before the 3-D reconstruction and two filters (Savitsky–Golay and Butterworth filters) after reconstruction were successively applied to minimize the effect of the 2-D occlusions. The filter parameters were optimized by minimizing the root mean square error between the reference and the filtered data. The optimal parameters of the filters were marker dependent, whereas no filter was necessary after a 2-D interpolation. As the occlusions cause systematic error in the 3-D reconstruction, the interpolation of the 2-D trajectories is more appropriate than filtering the 3-D trajectories.

Restricted access

Daniel J. Davis and John H. Challis

) Error time series for each of the different signal processing approaches along with the %RMSE, Absolute %Impact Peak Error, and Geers %Phase Error values for the specific estimate depicted graphically in panel (A). %RMSE indicates percent root mean square error; SSF-BS, signal-section filtering by body

Restricted access

Patrick Ippersiel, Richard Preuss, and Shawn M. Robbins

? J Psychiatr Res . 1996 ; 30 ( 6 ): 483 – 492 . PubMed ID: 9023792 doi:10.1016/S0022-3956(96)00033-7 9023792 10.1016/S0022-3956(96)00033-7 13. Klingspor M . Hilbert Transform: Mathematical Theory and Applications to Signal Processing [master’s thesis]. Linköping, Sweden : Linköping University

Restricted access

Joshua Twaites, Richard Everson, Joss Langford, and Melvyn Hillsdon

Purpose: Physical activity classifiers are typically trained on data obtained from sensors at a set orientation. Changes in this orientation (such as being on a different wrist) result in performance degradation. This work investigates a method to obtain sensor location and orientation invariance for classification of wrist-mounted accelerometry via a technique known as domain adaption. Methods: Data was gathered from 16 participants who wore accelerometers on both wrists. Physical activity classification models were created using data from each wrist and then used to predict activities when using data from the opposing wrist. Using subspace alignment domain adaption, this procedure was then repeated to align the training and testing data before the classification stage. Results: Prediction of activity when using data where the wearer’s wrist was incorrectly specified resulted in a significant (p = .01) decrease in performance of 12%. When using domain adaption this drop in performance became negligible (M difference < 1%, p = .73). Conclusion: Domain adaption is a valuable method for achieving accurate physical activity classification independent of sensor orientation in wrist-worn accelerometry.

Restricted access

Francisco Javier Alonso, Publio Pintado, and José María Del Castillo

The use of the Hodrick-Prescott (HP) filter is presented as an alternative to the traditional digital filtering and spline smoothing methods currently used in biomechanics. In econometrics, HP filtering is a standard tool used to decompose a macroeconomic time series into a nonstationary trend component and a stationary residual component. The use of the HP filter in the present work is based on reasonable assumptions about the jerk and noise components of the raw displacement signal. Its applicability was tested on 4 kinematic signals with different characteristics. Two are well known signals taken from the literature on biomechanical signal filtering, and the other two were acquired with our own motion capture system. The criterion for the selection of cutoff frequency was based on the power spectral density of the raw displacement signals. The results showed the technique to be well suited to filtering biomechanical displacement signals in order to obtain accurate higher derivatives in a simple and systematic way. Namely, the HP filter and the generalized cross-validated quintic spline (GCVSPL) produce similar RMS errors on the first (0.1063 vs. 0.1024 m/s2) and second (23.76 vs. 23.24 rad/s2) signals. The HP filter performs slightly better than GCVSPL on the third (0.209 vs. 0.236 m/s2) and fourth (1.596 vs. 2.315 m/s2) signals.

Restricted access

Bernd J. Stetter, Erica Buckeridge, Vinzenz von Tscharner, Sandro R. Nigg, and Benno M. Nigg

This study presents a new approach for automated identification of ice hockey skating strides and a method to detect ice contact and swing phases of individual strides by quantifying vibrations in 3D acceleration data during the blade–ice interaction. The strides of a 30-m forward sprinting task, performed by 6 ice hockey players, were evaluated using a 3D accelerometer fixed to a hockey skate. Synchronized plantar pressure data were recorded as reference data. To determine the accuracy of the new method on a range of forward stride patterns for temporal skating events, estimated contact times and stride times for a sequence of 5 consecutive strides was validated. Bland-Altman limits of agreement (95%) between accelerometer and plantar pressure derived data were less than 0.019 s. Mean differences between the 2 capture methods were shown to be less than 1 ms for contact and stride time. These results demonstrate the validity of the novel approach to determine strides, ice contact, and swing phases during ice hockey skating. This technology is accurate, simple, effective, and allows for in-field ice hockey testing.

Restricted access

Joshua Twaites, Richard Everson, Joss Langford, and Melvyn Hillsdon

Introduction: Data from wrist-worn accelerometers often has an inherent natural segmentation that reflects transitioning from one activity to another. The aim of this study was to develop an activity transition detection method to realize this natural segmentation. Methods: Data was gathered from 16 participants who wore triaxial wrist accelerometers in a lab-based protocol and 47 participants in a free-living protocol. Change point detection was used to create a method for detecting activity transitions. The agreement between observed and predicted transitions was assessed by the Matthews Correlation Coefficient (MCC), Root Mean Squared Error (RMSE), and two additional metrics created for this task; the Ratio of Minimum Mean Distance (RMMD) and the Ratio of Sensitivity (RoS). The effects of varying combinations of acceleration axes were also investigated to determine the most effective set of axes. A novel post-processing technique was developed to mitigate a major limitation identified in current transition detection methods. Results: The developed transition detection method achieved a MCC of 0.763, a RMSE of 3.17, a RoS of 2.40, and a RMMD of 3.21, outperforming existing techniques. The post-processing technique developed improved the performance of all methods when identifying transitions. It was found that using solely the y-axis (vertical acceleration) allowed for optimal performance. Conclusion: Change point detection is a valid method for identifying transitions in activity using wrist-worn accelerometer data. The new post processing technique developed improves the performance of transition detection methods.

Restricted access

Bing Yu, David Gabriel, Larry Noble, and Kai-Nan An

The purposes of this study were (a) to develop a procedure for objectively determining the optimum cutoff frequency for the Butterworth low-pass digital filler, and (b) to evaluate the cutoff frequencies derived from the residual analysis. A set of knee flexion-extension angle data in normal gait was used as the standard data set. The standard data were sampled at different sampling frequencies. Random errors with different magnitudes were added to the standard data to create different sets of raw data with a given sampling frequency. Each raw data set was filtered through a Butterworth low-pass digital filter at different cutoff frequencies. The cutoff frequency corresponding to the minimum error in the second time derivatives for a given set of raw data was considered as the optimum for that set of raw data. A procedure for estimating the optimum cutoff frequency from the sampling frequency and estimated relative mean error in the raw data set was developed. The estimated optimum cutoff frequency significantly correlated to the true optimum cutoff frequency with a correlation determinant value of 0.96. This procedure was applied to estimate the optimum cutoff frequency for another set of kinematic data. The calculated accelerations of the filtered data essentially matched the measured acceleration curve. There is no correlation between the cutoff frequency derived from the residual analysis and the true optimum cutoff frequency. The cutoff frequencies derived from the residual analysis were significantly lower than the optimum, especially when the sampling frequency is high.

Open access

Ignacio Perez-Pozuelo, Thomas White, Kate Westgate, Katrien Wijndaele, Nicholas J. Wareham, and Soren Brage

Background: Wrist-worn accelerometry is the commonest objective method for measuring physical activity in large-scale epidemiological studies. Research-grade devices capture raw triaxial acceleration which, in addition to quantifying movement, facilitates assessment of orientation relative to gravity. No population-based study has yet described the interrelationship and variation of these features by time and personal characteristics. Methods: 2,043 United Kingdom adults (35–65 years) wore an accelerometer on the non-dominant wrist and a chest-mounted combined heart-rate-and-movement sensor for 7 days free-living. From raw (60 Hz) wrist acceleration, we derived movement (non-gravity acceleration) and pitch and roll (forearm) angles relative to gravity. We inferred physical activity energy expenditure (PAEE) from combined sensing and sedentary time from approximate horizontal arm angle coupled with low movement. Results: Movement differences by time-of-day and day-of-week were associated with forearm angles; more movement in downward forearm positions. Mean (SD) movement was similar between sexes ∼31 (42) mg, despite higher PAEE in men. Women spent longer with the forearm pitched >0°, above horizontal (53% vs 36%), and less time at <0° (37% vs 53%). Diurnal pitch was 2.5–5° above and 0–7.5°below horizontal during night and daytime, respectively; corresponding roll angles were ∼0° (hand flat) and ∼20° (thumb-up). Differences were more pronounced in younger participants. All diurnal profiles indicated later wake-times on weekends. Daytime pitch was closer to horizontal on weekdays; roll was similar. Sedentary time was higher (17 vs 15 hours/day) in obese vs normal-weight individuals. Conclusions: More movement occurred in forearm positions below horizontal, commensurate with activities including walking. Findings suggest time-specific population differences in behaviors by age, sex, and BMI.

Restricted access

Dinesh John, Qu Tang, Fahd Albinali, and Stephen Intille

propose a method that first uses digital signal processing techniques to harmonize raw data from devices with different dynamic range and sampling rates, and then aggregates the raw data to yield a M onitor- I ndependent M ovement S ummary unit (MIMS-unit) that captures normal human motion. We outline