Figure 3 - uploaded by J.J. Little
Content may be subject to copyright.
Sample image, number 39 from a sequence of 78, of an experimental subject walking across the field-of-view of the camera. Images were initially full-color, 640 by 480 pixels. They were re-sampled and cropped to 320 by 160 pixels.
Source publication
Moving light displays (MLDs) have been used extensively to study motion perception and perception of the human gait in particular. MLD perception is largely considered to be structural, i.e., perception depends on identification of human kinematic structure. However, work by Little and Boyd (1996) has shown that it is possible to recognize individu...
Citations
... Using the characteristics of this new feature descriptor, we utilize the l 2 -regularized collaborative representation classifier for human action recognition. In certain cases, this special classifier could well perform on human action recognition [9]. ...
... The actions are partitioned into three subsets as shown in Table 1. A total of 20 actions are employed and half of the total subjects (1,3,5,7,9) are utilized for training and the rest ones for testing. ...
... However, the 128 × 128 sizes showed better results globally, thus we chose 128 × 128 as a standard size in all our experiments. (10) Draw circle (9) Tennis swing (17) Bend (13) Two hand wave (11) Tennis serve (18) Tennis serve (18) Forward kick (14) Golf swing (19) Pickup throw (20) Side boxing (12) Pickup throw (20) As for the five selected sub-bands, they were chosen from different decomposition level conducted an experiment by selecting sub from the same decomposition level, but the recognition accuracy is only promising locally, not globally. The experimental results showed that the combination of the high-frequency sub bands on the first level and the low sub-band on the last level, always obtains promising results for all test cases with 128 × 128 DMMs. ...
The emerging cost-effective depth sensors have made easier the action recognition task significantly. In this paper, we propose an effective method to analysis human actions from depth video sequences based on multi-scaling and multi-directional transformation which provide additional body shape and motion information for action recognition. In our method, corresponding to the front, side and top projection views, we generate three Depth Motion Maps (DMMs) over the entire video sequences. More specially, the multi-scaling and multi-directional transformations are implemented on the generated DMMs of a depth video sequence. Finally, the concatenation of these features is used as a feature descriptor for the depth video sequence. With this new feature descriptor, the l2-regularized collaborative representation classifier (l2- CRC) is utilized to recognize human actions. The recognition results of Microsoft Research (MSR) Action3D dataset show that our method significantly outperforms than the other existing methods, although our representation is much more compact.
... Another large category of approaches extracts cues about the activity taking place from motion information [12]. One such approach examines the global shape of motion features, which are found to provide enough information for recognition [13]. The periodicity of human motions is used in [14] to derive templates for each action class, but at a high computational cost, as it is based on the correlation of successive video frames. ...
The widespread use of digital multimedia in applications, such as security, surveillance, and the semantic web, has made the automated characterization of human activity necessary. In this work, a method for the characterization of multiple human activities based on statistical processing of the video data is presented. First the active pixels of the video are detected, resulting in a binary mask called the Activity Area. Sequential change detection is then applied to the data examined in order to detect at which time instants there are changes in the activity taking place. This leads to the separation of the video sequence into segments with different activities. The change times are examined for periodicity or repetitiveness in the human actions. The Activity Areas and their temporal weighted versions, the Activity History Areas, for the extracted subsequences are used for activity recognition. Experiments with a wide range of indoors and outdoors videos of various human motions, including challenging videos with dynamic backgrounds, demonstrate the proposed system's good performance.
... All model-based approaches, however, are faced with the challenge of matching model parameters of varying complexity to a human image. Nonmodel-based systems, on the other hand, recognize human activity by nonstructural means using global shape of motion features [19]. Periodicity of human locomotion is one such motion feature that has often been used as a recognition criterion. ...
Recent advances in computer vision and pattern recognition have fueled numerous initiatives that aim to intelligently recognize human activities. In this paper, we propose an algorithm for nonintrusive human activity recognition. We use an adaptive background-foreground separation technique to extract motion information and generate silhouettes (foreground) from the input videos. We then derive directionality-based feature vectors (directional vectors) from the silhouette contours and use the distinct data distribution of directional vectors in a vector space for clustering and recognition. We also exploit the dynamic characteristic of human motion in order to smooth decisions over time and reduce errors in activity recognition. Our approach is monocular, tolerant to moderate view changes, and can be applied to both frontal and lateral views of most activities. Experiments with short and long video sequences show robust recognition under conditions of varying view angles, zoom depths, backgrounds, and frame rates.
... In the case of machine vision, the biological metaphor suggests that it may be possible to use reduced spatio-temporal information, such as embedded in MLDs, for recognition. MLDs images, as feature-based motion cues, have been widely used in studies of visual perception [3,12,20,24,28]; human motion tracking and activity recognition in computer vision678913,19]; clinical gait analysis and sports science research [2,14,37,38,43]; character animation [18,35] ; augmented reality and virtual reality [13]. Motion analysis from reduced MLDs allows us to use quantitative, concise and accurate data to investigate essential recognition features in visual perception, motion modelling, kinematic formulation and motion synthesis. ...
... Though relatively few researchers have attempted motion-based recognition, Abdelkader et al. [1] proposed a motion-based structure-free method to characterise motion pattern in monocular video for human gait recognition. Boyd and Little [6], using global shape-of-motion features derived from MLD images, has shown that it is possible to recognise individual people by their gait using non-structural means. Recent work by Wang et al. [41,42], employing spatial– temporal silhouette as biometric motion signature and PACbased eigenvalue analysis, achieved successful gait recognition from outdoor image sequences in a reduced dimensionality of feature space. ...
... Approaches using motion directly, without regard to its underlying structure, for (human) periodic motion recognition are described in e.g. [1,4,6,10,15,32,33,39,41,42] . Motionbased approaches characterise human periodic motion by, for example, a set of static configurations of the body in each pose in a manner of state-spaces, or by analysing shape of motion, trajectories, templates and optical flow images in spatial– temporal dimensions simultaneously. ...
Feature-based motion cues play an important role in biological visual perception. We present a motion-based frequency-domain scheme for human periodic motion recognition. As a baseline study of feature based recognition we use unstructured feature-point kinematic data obtained directly from a marker-based optical motion capture (MoCap) system, rather than accommodate bootstrapping from the low-level image processing of feature detection. Motion power spectral analysis is applied to a set of unidentified trajectories of feature points representing whole body kinematics. Feature power vectors are extracted from motion power spectra and mapped to a low dimensionality of feature space as motion templates that offer frequency domain signatures to characterise different periodic motions. Recognition of a new instance of periodic motion against pre-stored motion templates is carried out by seeking best motion power spectral similarity. We test this method through nine examples of human periodic motion using MoCap data. The recognition results demonstrate that feature-based spectral analysis allows classification of periodic motions from low-level, un-structured interpretation without recovering underlying kinematics. Contrasting with common structure-based spatio-temporal approaches, this motion-based frequency-domain method avoids a time-consuming recovery of underlying kinematic structures in visual analysis and largely reduces the parameter domain in the presence of human motion irregularities.
... Global representations are the lowest level of representation which captures the whole body motion. Structured representations are a higher-level representation that may record only the motion of specific structural components of the human body (Boyd and Little 1997). A structured representation requires tracking of specific body parts (joints) of the actor while simplifies the classification process involved in movement recognition. ...
1 The support of ARDA (VACE), NSF (HSD), and CAPES is gratefully acknowledged. Abstract We present a roadmap to a Human Activity Language (HAL) for symbolic manipulation of visual and motor information in a sensory-motor system model. The visual perception subsystem translates a visual representation of action into our visuo-motor language. One instance of this perception process could be achieved by a Motion Capture system. We captured almost 90 different human actions in order to have empirical data that could validate and support our embodied language for movement and activity. The embodiment of the language serves as the interface between visual perception and the motor subsystem. The visuo-motor language is defined using a linguistic approach. In phonology, we define basic atomic segments that are used to compose human activity. Phonological rules are modeled as a finite automaton. In morphology, we study how visuo-motor phonemes are combined to form strings representing human activity and to generate a higher-level morphological grammar. This compact grammar suggests the existence of lexical units working as visuo-motor subprograms. In syntax, we present a model for visuo-motor sentence construction where the subject corresponds to the active joints (noun) modified by a posture (adjective). A verbal phrase involves the representation of the human activity (verb) and timing coordination among different joints (adverb).
... Human motion analysis has gained increasing attention from computer vision researchers motivated by a wide spectrum of applications such as surveillance, medical, man–machine interface and animation. The major areas of research are motion analysis [8,9], tracking [10,11], recognizing biological motion [12,13] , and now as a biometric . Investigations into gait as a biometric only began about a decade ago. ...
Gait enjoys advantages over other biometrics in that it can be perceived from a distance and is difficult to disguise. Current approaches are mostly statistical and concentrate on walking only. By analysing leg motion we show how we can recognise people not only by the walking gait, but also by the running gait. This is achieved by either of two new modelling approaches which employ coupled oscillators and the biomechanics of human locomotion as the underlying concepts. These models give a plausible method for data reduction by providing estimates of the inclination of the thigh and of the leg, from the image data. Both approaches derive a phase-weighted Fourier description gait signature by automated non-invasive means. One approach is completely automated whereas the other requires specification of a single parameter to distinguish between walking and running. Results show that both gaits are potential biometrics, with running being more potent. By its basis in evidence gathering, this new technique can tolerate noise and low resolution.
... This approach based on template matching [22,82,161,108,109,147], first converts an image sequence into a static shape pattern, and then compares it to prestored action prototypes during recognition. In the early work by Polana and Nelson [82], the features consisting of two-dimensional meshes were utilized to recognize human action. ...
Visual analysis of human motion is currently one of the most active research topics in computer vision. This strong interest is driven by a wide spectrum of promising applications in many areas such as virtual reality, smart surveillance, perceptual interface, etc. Human motion analysis concerns the detection, tracking and recognition of people, and more generally, the understanding of human behaviors, from image sequences involving humans. This paper provides a comprehensive survey of research on computer-vision-based human motion analysis. The emphasis is on three major issues involved in a general human motion analysis system, namely human detection, tracking and activity understanding. Various methods for each issue are discussed in order to examine the state of the art. Finally, some research challenges and future directions are discussed.
... Movements of objects and their movement behaviors repeat periodically in space (x, y ) and time (t). Identifying repeated movements and behaviors of an object is a strong cue for object and movement recognition (Boyd and Little, 1997;Goddard, 1997). ...
Several applications generate large volumes of data on movements including vehicle navigation, fleet management, wildlife tracking and in the near future cell phone tracking. Such applications require support to manage the growing volumes of movement data. Understanding how an object moves in space and time is fundamental to the development of an appropriate movement model of the object. Many objects are dynamic and their positions change with time. The ability to reason about the changing positions of moving objects over time thus becomes crucial. Explanations on movements of an object require descriptions of the patterns they exhibit over space and time. Every moving object exhibits a wide range of patterns some of which repeat but not exactly over space and time such as an animal foraging or a delivery truck moving about a city. Even though movement patterns are not exactly the same, they are not completely different. Moving objects may move on the same or nearly similar paths and visit the same locations over time. This thesis addresses the identification of repeat movement patterns from large volumes of data. These are represented as higher-level movement structures referred to as movement signatures. Movement signatures are defined as collections of patterns that objects demonstrate in their sequences of movements. Signatures have a spatial structure that includes dominant or frequently visited locations and paths and a spatio-temporal structure that associates a temporal pattern with the spatial patterns. This thesis demonstrates the extraction of movement signatures from sets of movement observations using fuzzy and Neuro-fuzzy methodologies. Identification of movement signatures and definition of their attributes provides summary level information for modeling and reasoning about moving objects.
... There is a substantial literature on the problem of tracking shapes in a sequence, for example the tracking of humans has recently been admirably surveyed [1]. Earlier research has ranged from optical ow [2] to Kalman ÿlters [3] and includes temporal templates , which we use here in an adaptation of the Hough Transform (HT) [4] to enable optimal extraction of moving arbitrary shapes. Temporal templates are a technique for representing the movement of bodies through a sequence of images by encoding a motion trajectory. ...
Many approaches can track objects moving in sequences of images but can suffer in occlusion and noise, and often require initialisation. These factors can be handled by techniques that extract objects from image sequences, especially when phrased in terms of evidence gathering. Since the template approach is proven for arbitrary shapes, we re-deploy it for moving arbitrary shapes, but in a way aimed to avoid discretisation problems. In this way, the discrete mapping operation is deferred as far as possible, by using continuous shape descriptions. A further advantage is reduction in computational demand, as seen in use of templates for shape extraction. This prior specification of motion avoids the need to use an expensive parametric model to capture data that is already known. Furthermore, the complexity of the motion template model remains unchanged with increase in the complexity of motion, whereas a parametric model would require increasingly more parameters leading to an enormous increase in computational requirements. The new approach combining moving arbitrary shape description with motion templates permits us to achieve the objective of low dimensionality extraction of arbitrarily moving arbitrary shapes with performance advantage as reflected by the results this new technique can achieve.
... Then they utilize both posture and motion cycle of the "star" skeleton to recognize activities such as walking and running. Recent work by Boyd and Little [2], using global shape-of-motion features derived from MLD images, has shown that it is possible to recognize individual people from their gaits, by non-structural means. ...
We present a frequency domain analysis technique for modelling and recognizing human periodic movements from moving light displays (MLDs). We model periodic motions by motion templates, that consist of a set of feature power vectors extracted from unidentified vertical component trajectories of feature points. Motion recognition is carried out in the frequency domain, by comparing an observed motion template with pre-stored templates. This method contrasts with common spatio-temporal approaches. The proposed method is demonstrated by some examples of human periodic motion recognition in MLDs.
