Vlasta Zanchi

Hand shape classification using DTW and LCSS as similarity measures for vision-based gesture recognition system

In this paper an approach to classify hand shapes into different classes according to the similarity measures between features is proposed. We show how to use an Exploratory Data Analysis to extract novel, single feature of hand from images. Based on the obtained curve-like shape of the feature, hands are classified into one of 21 possible classes of Croatian sign language using Dynamic Time Warping and Longest Common Subsequence as similarity measures. Performance of the system was evaluated with 1260 images. Results show that high classification accuracy can be obtained from a single feature recognition and a small number of training sample.

Quantitative human gait analysis

Abstract In this paper, the methodology for normal gait recognition and estimation is described. Normal gait recognition is derived on the basis of kinematics data of the human locomotion system. Measurements were carried out and the data were processed and statistically analyzed. The procedure was done on a group of 20 students. Kinematics data have been presented in phase plane. Sets of data in phase plane for the specific discrete moments in time were statistically processed using the Gaussian and Bootstrap methods. Discrete moments are chosen according to specific gait phases of a gait cycle. Finally, as a result of statistical analysis, the gait quality index (GQI) is obtained for each gait phase.

Motion analysis system for identification of 3D human locomotion kinematics data and accuracy testing

Abstract In this article, a procedure for acquisition and processing of the human locomotion system kinematics data is presented. Software for 3D motion analysis system is developed, discussed and described. Data acquisition was performed using two commercial camcorders, framegrabber and a personal computer. Data processing by direct linear transformation provided 3D coordinates of the measured movement. Inaccuracies due to the relatively low frame rate and non-synchronised cameras were improved using software. The results of laboratory testing are given, and confirm this improvement. The procedure for synchronisation error elimination is described in detail.

Is it possible to detect mobile phone user's attention based on accelerometer measurment of gait pattern?

Mobile phones have become ubiquitous in todays world. Their ever increasing computational power and sensing capabilities have made them well suited for number of tasks well beyond their original purpose of communication. But mobile phone usage while walking or driving can potentially be dangerous leading to serious injury or even death. In the paper we answer the question is it possible using only mobile phones embedded accelerometer to detect changes in gait pattern caused by changed attention level due to interaction with mobile device like reading on-screen text. Experimental measurements were conducted on 8 test subjects in indoor environment with each test subject performing 6 trials. Two different approaches based on gait phase and gait velocity were tested on recorded data in batch mode with more promising one implemented in real-time manner. Obtained results are presented and discussed and possible future research directions outlined.

LMNN metric learning and fuzzy nearest neighbour classifier for hand gesture recognition

This paper presents a novel gesture recognition system using a single three-axis accelerometer, that is to serve as an alternative or supplementary interaction modality for controlling mobile devices. Capturing, training and classification of the detected hand gestures are expected to be executed in their entirety on the mobile device running the proposed system, instead of being passed to a nearby computer. As gesture recognition belongs to the group of pattern recognition problems where the underlying class probabilities are not a priori known, the classification is based on the distance between neighbouring examples. The distance metric is optimized by using large margin nearest neighbour (LMNN) method. To measure the amount of classification confidence, a fuzzy version of nearest neighbour algorithm is employed. Obtained results for recognition of nine hand gestures using proposed LMNN—fuzzy combination are presented and compared to that of other similar approaches. The system achieves near perfect recognition accuracy that is highly competitive with systems based on statistical methods and other accelerometer-based gesture recognition systems in the literature.

Object Recognition Techniques in Behavioral Analysis

In this paper we present two methods for automatic quantification of TV consumer attention. The problem has been divided into two main parts: the first one is oriented toward appropriate image analysing techniques which are problem specific and include superposition of several well defined technology for object recognition, movement analysis and tracking, whereas the second part includes more sociological aspects of image analysis interpretation in the context of typical human (kids) behaviour during their, unfortunately time consuming, usual daily activities.

Color cues in Human Motion Analysis

An important part in the analysis of human activity video data is the silhouette segmentation. The results of segmentation are greatly affected by imaging environment, thus posing a problem for the extraction of the region in the image containing a subject of interest. In this paper we propose a human motion analysis method based on nonparametric clustering of monocular color images. Motivated by the need to automatically extract human silhouettes for kinematic gait analysis, without spurious segmentations naturally occurring in other segmentation methods, we have developed and applied adaptive mesh-based color clustering which can be combined with motion segmentation. The advantage of our method is that it is controlled by few intuitive parameters allowing the method to be adjusted to different capturing environments. The usage of color is not sensitive to illumination variations, or to different color distributions of different cameras, as the color distribution is compared between different regions of the same frame. To provide a qualitative evaluation of our method, our results are compared with the Gaussian Mixture model, a standard vision-based human subject extraction method based on background segmentation techniques.

Estimation of hand preshaping during human grasping by using instrumented glove

This paper deals with the novel method for evaluating hand preshaping during reaching-to-grasp movement. The method makes use of all five fingers in estimation of prehension [1]. The investigation was performed on six healthy subjects grasping three different objects at various positions and orientations. The objects were presented to the subjects by means of a robot, which also induced perturbations in both object position and orientation. Positions of markers attached to the finger-tips and dorsum of the hand were recorded by means of a 3D optical tracking system, Optotrak. In the data analysis the adjacent finger-tips were interconnected, thus obtaining a planar pentagon whose various characteristics were investigated and discussed. New parameters for the evaluation of finger preshaping, such as pentagon surface area, angle between the pentagon and hand normal vectors, and the angle between the pentagon and object normal vectors between were introduced, [1]. We have also investigated the hand orientation during prehensile movements by analysing the angle between the vectors emanating from the palm and object to be grasped [2]. As the next step of improving the applicability of the method is the implementation of the instrumented glove. 3D optical motion tracking system is reliable, but robust and complicate to use, especially in the rehabilitation environment. On the other hand, the instrumented glove is easy to use, and, as we shoved, provides reliable data. We performed simultaneous measurement of prehensile movement by glove and Optotrak. We used biomechanical model of the hand to calculate the positions of the fingertips on the basis of the fingers joint angles measured by the glove. The data calculated from the glove outputs were compared with the markers positions recorded by Optotrak. So obtained data were used to calculate hand opening described by pentagon square area. The proposed pentagon approach is expected to be useful in future work when examining grasping abilities of subjects with neuromuscular disorders.