Publications (4)0 Total impact
-
[show abstract]
[hide abstract]
ABSTRACT: We present an algorithm for quantity motion capture and multi camera HD 1080 standard reference video data fusion. It consists
of initial calibration step which is based on some set of selected frames and final fusion for the rest of frames. Implemented
data fusion algorithm can be used in case that it is possible to find a time interval when both devices were recording the
same sequence of poses. It is worth to emphasise there are no special calibration patterns used during calibration. Advantage
of the algorithm is that the required calibration step can be perfomed simultaneously with actor calibration from Vicon Blade
system. It is also allowed that cameras locations can be changed during acquisition process as long as they observe known
motion capture markers. After calibration and synchronization reprojection is possible in real time for VGA resolution or
in reduced frequency for HD 1080 standard. Performed experiments determined that average projection error is about 1.45 pixel
in the Full-HD 1920×1080 reference video and it is perceptualy acceptable. Practical usage for training video depersonification
was presented.
08/2011: pages 209-216;
-
[show abstract]
[hide abstract]
ABSTRACT: Gait paths are spatial trajectories of selected body points during person’s walk. We have proposed and evaluated features
extracted from gait paths for the task of person identification. We have used the following gait paths: skeleton root element,
feet, hands and head. In our motion capture laboratory we have collected human gait database containing 353 different motions
of 25 actors. We have proposed four approaches to extract features from motion clips: statistical, histogram, Fourier transform
and timeline We have prepared motion filters to reduce the impact of the actor’s location and actor’s height on the gait path.
We have applied supervised machine learning techniques to classify gaits described by the proposed feature sets. We have prepared
scenarios of the features selections for every approach and iterated classification experiments. On the basis of obtained
classifications results we have discovered most remarkable features for the identification task. We have achieved almost 97%
identification accuracy for normalized paths.
Keywordsmotion capture–human identification–gait recognition–supervised learning–features extraction–features selection–biometrics
08/2011: pages 531-542;
-
[show abstract]
[hide abstract]
ABSTRACT: We have focused on the problem of classification of motion frames representing different poses by supervised machine learning
and dimensionality reduction techniques. We have extracted motion frames from global database manually, divided them into
six different classes and applied classifiers to automatic pose type detection. We have used statistical Bayes, neural network,
random forest and Kernel PCA classifiers with wide range of their parameters. We have tried classification on the original
data frames and additional reduced their dimensionality by PCA and Kernel PCA methods. We have obtained satisfactory results
rated in best case 100 percent of classifiers efficiency.
09/2010: pages 193-200;
-
[show abstract]
[hide abstract]
ABSTRACT: In this paper we present a simple estimation system for ground reaction forces, inter-segmental forces and joint torques in
the biped motion. The proposed system bases on classification of walk into two phases, left leg stance - right leg swing,
and left leg swing - right leg stance. The system does not need any additional measurements of forces and moments. It uses
matrix formulation for inverse dynamics problem of computing moment and torques on the basis of measurements of ve- locities
and accelerations. The input signals are filtered with the use of the Savitzky Golay smoothing filter.
09/2010: pages 185-192;
