Joint center location is the driving parameter for determining the kinematics, and later kinetics, associated with human motion capture. Therefore the accuracy with which said location is determined is of great import to any and all subsequent calculation and analysis. The most significant barrier to accurate determination of this parameter is soft tissue artifact, which contaminates the measurements of on-body measurement devices by allowing them to move relative to the underlying rigid bone. This leads to inaccuracy in both bone pose estimation and joint center location. The complexity of soft tissue artifact (it is nonlinear, multimodal, subject-specific, and trial specific) makes it difficult to model, and therefore difficult to mitigate. This thesis proposes a novel method, termed Single Frame Optimization, for determining joint center location (though mitigation of soft tissue artifact) via a linearization approach, in which the optimal vector relating a joint center to a corresponding inertial sensor is calculated at each time frame. This results in a time-varying joint center location vector that captures the relative motion due to soft tissue artifact, from which the relative motion could be isolated and removed. The method’s, and therefore the optimization’s, driving assumption is that the derivative terms in the kinematic equation are negligible relative to the rigid terms. More plainly, it is assumed that any relative motion can be assumed negligible in comparison with the rigid body motion in the chosen data frame. The validity of this assumption is investigated in a series of numerical simulations and experimental investigations. Each item in said series is presented as a chapter in this thesis, but retains the format of a standalone article. This is intended to foment critical analysis of the method at each stage in its development, rather than solely in its practical (and more developed) form.

Modern soldiers are burdened by an increase in body-borne load due to technological advancements related to their armour and equipment. Despite the potential increase in safety from carrying more protective equipment, a heavier load on the soldier might decrease field performance both cognitively and physically. Additionally, an increasing load on military personnel concurrently increases their risk of musculoskeletal injuries. Therefore, there is a necessity for research on the soldier's biomechanical outcomes under different loading conditions. When it comes to biomechanics research, marker-based technology is widely accepted as the gold standard in terms of motion capture. However, recent advancements in markerless motion capture could allow the quick collection of data in various training environments, while avoiding marker errors. In this research project, the Theia3D markerless motion capture system was compared to the marker-based gold standard for application on participants across varying body-borne load conditions. The aim was to estimate lower body joint kinematics, gastrocnemius lateralis and medialis muscle activation patterns, and lower body joint reaction forces from the two motion capture systems. Data were collected on 16 participants for three repetitions of both walking and running under four body-borne load conditions by both motion capture systems simultaneously. Electromyography (EMG) data of lower limb muscles were collected on the right leg and force plates measured ground reaction forces. A complete musculoskeletal analysis was completed in OpenSim using the Rajagopal full-body model and standard workflow: model scaling, inverse kinematics, residual reduction, static optimization, and joint reaction analysis. Estimations of joint kinematics and joint reaction forces were compared between the two systems using Pearson's correlation coefficient, root-mean-square errors, and Bland-Altman limits of agreement. Very strong correlations (r = 0.960 ± 0.038) and acceptable differences (RMSE = 7.8° ± 2.6°) were observed between the kinematics of the marker-based and markerless systems, with some angle biases due to joint centre differences between systems causing an offset. Because the marker-based motion capture system lost line of sight with markers more frequently in the heavier body-borne load conditions, differences generally increased with heavier body-borne loads. Timing of muscle activations of the gastrocnemius lateralis and medialis as estimated from both systems agreed with the ones measured by the EMG sensors. Joint reaction force results also showed a very strong correlation between the systems but the markerless model seemed to overestimate joint reaction forces when compared to results from the marker-based model. Overall, this research highlighted the potential of markerless motion capture to track participants across all body-borne load conditions. However, more work is necessary on the determination of angle bias between the two systems to improve the use of markerless data with OpenSim models.

Motion capture(“MoCap”) has been used for a long time in the movie and videogame industries to animate digital characters. This technology commonly requires a studio and expensive stationary equipment. However, in recent years markerless MoCap has emerged. This is a technology that uses machine learning to estimate and reproduce the movements of humans. This technology can be used with a single video camera thus making it more accessible. This study relates to research on motion capture, machine learning and computer animation. The study examines a selection of markerless MoCap tools available on the market with amateurs and small businesses as target audiences. This to explore to which extent markerless MoCap for amateurs is suitable for use. The research questions asked in this study are: How well do these tools recreate motions from an animation? How are these results affected by aggravating circumstances? How do the results of the tools differ from each other? To explore these questions, a selection of five markerless MoCap services was made. These five services were then tested to study their performances in different aggravating circumstances. An original animation was created and used in these tests. The results from these tests were analyzed using a qualitative visual analysis and a numerical analysis of extreme values. The study found the tools could not accurately reproduce the animation they were given to process. The most prominent problem being that of depth perception, which resulted in the processed animations often deviating in depth. The services also had obvious problems with recreating arms. The study also found that some of the different aggravating circumstances affected the results more than others. The results of this study shows that markerless MoCap for amateurs still has development ahead of it before the technology can be considered an effective tool. / Motion Capture (“MoCap”) har länge använts inom film- och spelindustrin för att animera digitala karaktärer. MoCap i storskalig produktion kräver dock vanligtvis en studio och dyr utrustning. Men på senare år har Markerless MoCap vuxit fram. Det är en teknik som använder sig av maskininlärning för att estimera och avbilda en persons rörelser. Denna kan användas med enbart en videokamera vilket gör tekniken lättillgänglig. Denna studie relaterar till forskning som berör Motion Capture, 3D-datoranimation, AI och maskininlärning. Studien undersöker ett urval av Markerless MoCap-verktyg som finns tillgängliga allmänheten, med amatörer och småföretag som målgrupp. Detta i syfte att undersöka i vilken utsträckning markerless MoCap för amatörer är lämplig för bruk. Problemformuleringen i denna studie är: Hur väl återskapar programmen rörelserna från en animation? Hur påverkas detta resultat av försvårande omständigheter? Hur skiljer sig dessa programs resultat från varandra? För att undersöka dessa frågor gjordes ett urval av fem Markerless MoCap-verktyg. Dessa fem verktyg testades för att studera verktygens prestationer under olika försvårande omständigheter. En egenproducerad animation användes i dessa tester. Resultaten från dessa tester analyserades med en kvalitativ visuell analys och en numerisk analys av extremvärden. Studien fann att verktygen inte med precision kunde återge den animation de fått att avbilda. Tydligast var problemet med djupseendet, vilket resulterade i att de bearbetade animationerna ofta avvek i djupled. Verktygen hade också påtagliga problem med att avbilda armar. Studien fann även att vissa försvårande omständigheter hade större effekt än andra. Den här studiens resultat visar att Markerless MoCap för amatörer fortfarande har utveckling kvar innan tekniken kan betraktas som ett effektivt verktyg.

No description available.

The latest advances of Virtual Reality technologies and three-dimensional graphics, as well as the developments in Gaming Technologies in the recent years, have stemmed the proliferation of Serious Games in a broader spectrum of research applications. Among the most popular areas of application are public services such as Defence and Health, where digital technologies realise new challenges and opportunities for research and development of Serious Games and for a variety of contexts. As with all games, the user engagement is elevated and apart from the entertaining aspect, Serious Games serve as a novel and promising alternative experience to knowledge transfer. Furthermore, Serious Games bring to the end user and the overall society a series of attractive benefits. These benefits include safety, cost-effectiveness, increased motivation and personalisation. Hence, this Thesis aims to investigate novel approaches of developing Serious Games that utilise the recent advances of Virtual Reality and Gaming Technology and facilitate the aforementioned benefits. The process of design and development of the novel tools and applications follow an iterative manner and are driven by the review of the available literature as well as end-user feedback.

The purpose of this paper is to examine and compare two simple motion capturesystems. The first system consists of two cameras, and the second system consists of acamera with a depth sensor. The problem with both systems is their limited field ofview. If the actor is facing a camera, that camera will not be able to see an arm if thatarm is placed behind the back.My method in order to do this examination, was to record a number of different typesof motions with both systems, and then evaluate and compare the generated data fromthe two systems. This was done partly by applying it to a 3D character, and partly byinvestigating how good incomplete data could be repaired. Some motions wasrecorded multiple times, each with a different angle. The examination has resulted indifferent conclusions about the two systems.The process of using two cameras was very time consuming and the actor had to standfacing the cameras more or less. The depth sensing camera was shown to be able tohandle a wider range of motions, and the angle relative to the cameras wasn't aslimited. Some advantages could be found compared to the depth sensing camera,which had troubles handling certain types of arm and head rotations. This waspossible to handle with two cameras. / Validerat; 20140627 (global_studentproject_submitter)

<p>Motion Capture is an essential part of a world full of digital effects in movies and games. Understanding the pipelines between software is a crucial component of this research. Methods that create the motion capture structure today are reviewed, and how they are implemented in order to create the movements that we see in modern games and movies.</p>

This work presents a novel technique for 2D human motion capture using a single non calibrated camera. The user's five extremities (head, hands and feet) are extracted, labelled and tracked after silhouette segmentation. As they are the minimal number of points that can be used in order to enable whole body gestural interaction, we will henceforth refer to these features as crucial points. The crucial point candidates are defined as the local maxima of the geodesic distance with respect to the center of gravity of the actor region which lie on the silhouette boundary. In order to disambiguate the selected crucial points into head, left and right foot, left and right hand classes, we propose a Bayesian method that combines a prior human model and the intensities of the tracked crucial points. Due to its low computational complexity, the system can run at real-time paces on standard Personal Computers, with an average error rate range between 2% and 7% in realistic situations, depending on the context and segmentation quality.

In this thesis, a system called Video Looping is developed to analyze human cyclic motions. Video Looping allows users to extract human cyclic motion from a given video sequence. This system analyzes similarities from a large amount of live footage to find the point of smooth transition. The final cyclic loop is created using only a few output images. Video Looping is useful not only to learn and understand human movements, but also to apply the cyclic loop to various artistic applications. To provide practical animation references, the output images are presented as photo plate sequences to visualize human cyclic motion similar to Eadweard Muybridge's image sequences. The final output images can be used to create experimental projects such as composited multiple video loops or small size of web animations. Furthermore, they can be imported into animation packages, and animators can create keyframe animations by tracing them in 3D software.

Near-fall detection is important for medical research since it can help doctors diagnose fall-related diseases and also help alert both doctors and patients of possible falls. However, in people’s daily life, there are lots of similarities between near-falls and other Activities of Daily Living (ADLs), which makes near-falls particularly difficult to detect. In order to find the subtle difference between ADLs and near-fall and accurately identify the latter, the movement of whole human body needs to be captured and displayed by a computer generated avatar. In this thesis, a wireless inertial motion capture system consisting of a central control host and ten sensor nodes is used to capture human body movements. Each of the ten sensor nodes in the system has a tri-axis accelerometer and a tri-axis gyroscope. They are attached to separate locations of a human body to record both angular and acceleration data with which body movements can be captured by applying Euler angle based algorithms, specifically, single rotation order algorithm and the optimal rotation order algorithm. According to the experiment results of capturing ten ADLs, both the single rotation order algorithm and the optimal rotation order algorithm can track normal human body movements without significantly distortion and the latter shows higher accuracy and lower data shifting. Compared to previous inertial systems with magnetometers, this system reduces hardware complexity and software computation while ensures a reasonable accuracy in capturing human body movements.