Search results

The Kinect sensor released by Microsoft is well-known for its effectiveness on human gesture recognition. Gesture recognition by Kinect has been proved to be an efficient command operation and provides an additional human-computer interface in addition to the traditional speech recognition. For Kinect gesture recognition in the application of gesture command operations, recognition of the active user making the gesture command to Kinect will be an extremely crucial problem. The purpose of this paper is to propose a recognition method for recognizing the person identity of an active user using combined eigenspace and Gaussian mixture model (GMM) with Kinect-extracted action gesture features.

Several Kinect-derived gesture features will be explored for determining the effective pattern features in the active user recognition task. In this work, a separate Kinect-derived feature design for eigenspace recognition and GMM classification is presented for achieving the optimal performance of each individual classifier. In addition to Kinect-extracted feature designs for active user recognition, this study will further develop a combined recognition method, called combined eigenspace-GMM, which properly hybridizes the decision information of both the eigenspace and the GMM for making a more reliable user recognition result.

Active user recognition using an effective combination of eigenspace and GMM with well-developed active gesture features in Kinect-based active user recognition will have an outstanding performance on the recognition accuracy. The presented Kinect-based user recognition system using the presented approach will further have the competitive benefits of recognition on both gesture commands and providing users of gesture commands.

A hybridized scheme of eigenspace and GMM performs better than eigenspace-alone or GMM-alone on recognition accuracy of active user recognition; a separate Kinect-derived feature design for eigenspace recognition and GMM classification is presented for achieving the optimal performance of the individual classifier; combined eigenspace-GMM active user recognition belonging to model-based active user recognition design has a fine extension on increasing the recognition rate by adjusting recognition models.

The aim of this research is to develop an eigenspace-based fuzzy c-means method for scalable topic detection.

The eigenspace-based fuzzy c-means (EFCM) combines representation learning and clustering. The textual data are transformed into a lower-dimensional eigenspace using truncated singular value decomposition. Fuzzy c-means is performed on the eigenspace to identify the centroids of each cluster. The topics are provided by transforming back the centroids into the nonnegative subspace of the original space. In this paper, we extend the EFCM method for scalability by using the two approaches, i.e. single-pass and online. We call the developed topic detection methods as oEFCM and spEFCM.

Our simulation shows that both oEFCM and spEFCM methods provide faster running times than EFCM for data sets that do not fit in memory. However, there is a decrease in the average coherence score. For both data sets that fit and do not fit into memory, the oEFCM method provides a tradeoff between running time and coherence score, which is better than spEFCM.

This research produces a scalable topic detection method. Besides this scalability capability, the developed method also provides a faster running time for the data set that fits in memory.

The authors propose a solder joint recognition method based on eigenspace technology.

The original solder joint image is transformed into a small set of feature subspace called “eigensolder”, which is the eigenvector of the training set and can represent a solder joint well. Then, the eigensolder feature is extracted by projecting the new solder joint image into the subspace, and the Euclidean distance measure is used to classify the solder joint.

The experimental results show that the proposed method is superior to the traditional classification method in solder joint recognition, and it can achieve 96.43 per cent recognition rate using only 15 eigenvalue images. It is suitable for the classification with small samples.

Traditional classification method like neural network and statistical method cost long time. Here, Eigensolder method is used to extract feature. Eigensolder method is more efficient, as it uses the principal component analysis method to reduce the feature dimension of input image and only measure the distance to classify.

The slow convergence of the incomplete Cholesky preconditioned conjugate gradient (CG) method, applied to solve the system representing a magnetostatic finite element model, is caused by the presence of a few little eigenvalues in the spectrum of the system matrix. The corresponding eigenvectors reflect large relative differences in permeability. A significant convergence improvement is achieved by supplying vectors that span approximately the partial eigenspace formed by the slowly converging eigenmodes, to a deflated version of the CG algorithm. The numerical experiments show that even roughly determined eigenvectors already bring a significant convergence improvement. The deflating technique is embedded in the simulation procedure for a permanent magnet DC machine.

From a theoretical point of view the question of spurious modes has been regarded as a closed problem. However, in this paper we show that even a precise definition of spurious‐free approximation was lacking. Hence, a sound definition of spurious‐free finite element method is given and a set of necessary and sufficient conditions ensuring that a finite element method is spurious‐free in the defined sense is stated. A critical comparison between the proposed theory and the currently accepted one is then carried out and existing counterexamples to the latter are pointed out. Comparison with an older theory leads to another set of necessary and sufficient conditions providing a better grasp of the key feature a finite element space must have to rule out spurious modes. The impact of the proposed theory is stressed, showing that Nedelec's tetrahedral edge elements of all orders provide spurious‐free approximations in all conditions of practical interest. Finally, it is shown, for the first time to the best of authors’ knowledge, that also many high‐order edge elements, recently proposed in the engineering literature for the analysis of electromagnetic problems, provide the same kind of reliable approximation.

Spurious modes often appear in the computed spectrum when an electromagnetic eigenproblem is solved by the finite element method. Demonstrates that the inclusion condition, often claimed as the theoretical reason for the absence of (non‐zero frequency) spurious modes, is a sufficient but not necessary condition for that. Does this by proving that edge elements, which are spectrally correct, do not satisfy the inclusion condition. As intermediate steps towards this result, proves the equivalence of the inclusion condition to a less cryptic one and gives two more easily‐checked necessary conditions for the latter. Concludes that from this investigation, the inclusion condition seems too strong to be useful as a sufficient condition. Works out the present analysis in the framework of spectral approximation theory for non‐compact operators, which emerges as a basic tool for a deeper understanding of the whole question of spurious modes.

Achieving natural interactions by means of vision and speech between humans and robots is one of the major goals that many researchers are working on. This paper aims to describe a gesture‐based human‐robot interaction (HRI) system using a knowledge‐based software platform.

A frame‐based knowledge model is defined for the gesture interpretation and HRI. In this knowledge model, necessary frames are defined for the known users, robots, poses, gestures and robot behaviors. First, the system identifies the user using the eigenface method. Then, face and hand poses are segmented from the camera frame buffer using the person's specific skin color information and classified by the subspace method.

The system is capable of recognizing static gestures comprised of the face and hand poses, and dynamic gestures of face in motion. The system combines computer vision and knowledge‐based approaches in order to improve the adaptability to different people.

Provides information on an experimental HRI system that has been implemented in the frame‐based software platform for agent and knowledge management using the AIBO entertainment robot, and this has been demonstrated to be useful and efficient within a limited situation.

Robot localization in dynamic, cluttered environments is a challenging problem because it is impractical to have enough knowledge to be able to accurately model the robot’s environment in such a manner. This study aims to develop a novel probabilistic method equipped with function approximation techniques which is able to appropriately model the data distribution in Markov localization by using the maximum statistical power, thereby making a sensibly accurate estimation of robot’s pose in extremely dynamic, cluttered indoors environments.

The parameter vector of the statistical model is in the form of positions of easily detectable artificial landmarks in omnidirectional images. First, using probabilistic principal component analysis, the most likely set of parameters of the environmental model are extracted from the sensor data set consisting of missing values. Next, we use these parameters to approximate a probability density function, using support vector regression that is able to calculate the robot’s pose vector in each state of the Markov localization. At the end, using this density function, a good approximation of conditional density associated with the observation model is made which leads to a sensibly accurate estimation of robot’s pose in extremely dynamic, cluttered indoors environment.

The authors validate their method in an indoor office environment with 34 unique artificial landmarks. Further, they show that the accuracy remains high, even when they significantly increase the dynamics of the environment. They also show that compared to those appearance-based localization methods that rely on image pixels, the proposed localization strategy is superior in terms of accuracy and speed of convergence to a global minima.

By using easily detectable, and rotation, scale invariant artificial landmarks and the maximum statistical power which is provided through the concept of missing data, the authors have succeeded in determining precise pose updates without requiring too many computational resources to analyze the omnidirectional images. In addition, the proposed approach significantly reduces the risk of getting stuck in a local minimum by eliminating the possibility of having similar states.

This study aims to model and investigate low-loss wave-propagation modes across random media. The objective is to achieve better channel properties for applying radio links through random vegetation (e.g. forest) using a beamforming approach. Thus, obtaining the link between the statistical parameters of the media and the channel properties.

A beamforming approach is used to obtain low-loss propagation across random media constructed of long cylinders, i.e. a simplified two dimensional (2D) model of agroforests. The statistical properties of the eigenmode radio wave propagation are studied following a Monte Carlo method. An error quantity is defined to represent the robustness of an eigenmode, and it is shown that it follows a known Lognormal statistical distribution, thereby providing a base for further statistical investigations.

In this study, it is shown that radio wave propagation eigenmodes exist based on a mathematical model. The algorithm presented can find such modes of propagation that are less affected by the statistical variation of the media than the regular beams used in radio wave communication techniques. It is illustrated that a sufficiently chosen eigenmode waveform is not significantly perturbed by the natural variation of the tree trunk diameters.

As a new approach to obtain low-loss propagation in random media at microwave frequencies, the presented mathematical model can calculate scattering-free wave-propagation eigenmodes. A robustness quantity is defined for a specific eigenmode, considering a 2D simplified statistical forest example. This new robustness quantity is useful for performing computationally low-cost optimization problems to find eigenmodes for more complex vegetation models.