Classification of natural hand gestures is usually approached by applying pattern recognition to the movements of the hand. However, the gesture categories most frequently cited in the psychology literature are fundam...
详细信息
ISBN:
(纸本)1581139543
Classification of natural hand gestures is usually approached by applying pattern recognition to the movements of the hand. However, the gesture categories most frequently cited in the psychology literature are fundamentally multimodal;the definitions make reference to the surrounding linguistic context. We address the question of whether gestures are naturally multimodal, or whether they can be classified from hand-movement data alone. First, we describe an empirical study showing that the removal of auditory information significantly impairs the ability of human raters to classify gestures. Then we present an automatic gesture classification system based solely on an n-gram model of linguistic context;the system is intended to supplement a visual classifier, but achieves 66% accuracy on a three-class classification problem on its own. This represents higher accuracy than human raters achieve when presented with the same information. Copyright 2004 ACM.
Two schemes for a charge pump are presented to reduce output ripple voltage with high load current drivability. The first is automatic pumping current control scheme which automatically adjusts the size of pumping dri...
详细信息
This paper proposes a new family of four degrees-of-freedom (dof) parallel mechanisms with two platforms and its application to a footpad device that can simulate the spatial motions of the human foot. The new mechani...
详细信息
This paper proposes a new family of four degrees-of-freedom (dof) parallel mechanisms with two platforms and its application to a footpad device that can simulate the spatial motions of the human foot. The new mechanism consists of front and rear platforms, and three limbs. Two limbs with 6-dof serial joints (Pe-S-P-P) are attached to each platform and are perpendicular to the base plate, while the middle limb (Pe-Re-R or P e-Pe-R) is attached to the revolute joint that connects the front and rear platforms. The middle limb is driven by the 2-dof driving mechanism that is equivalent active serial prismatic and revolute joints (P e-Re), or prismatic and prismatic joints (P e-Pe) with two base-fixed prismatic actuators. Therefore, two new 4-dof parallel mechanisms with two platforms can generate pitch motion of each platform, and roll and heave motions (1T-3R) or pitch motion of each platform and two translational motions (2T-2R) at both platforms. Kinematic analyses of the 1T-3R mechanism were performed, including inverse and forward kinematics, and velocity analysis. Based on the 1T-3R mechanism, a footpad device was designed to generate foot trajectories for natural walking. Finally, simulations of the foot trajectories in the normal gait cycle were performed using the proposed footpad device.
A study was done to evaluate weather comp.ter understanding of natural human speech can be improved by attending to non-verbal modalities. The extent to which gesture can supplement existing, purely verbal systems for...
详细信息
ISBN:
(纸本)1581139543
A study was done to evaluate weather comp.ter understanding of natural human speech can be improved by attending to non-verbal modalities. The extent to which gesture can supplement existing, purely verbal systems for speech understanding were investigated. The most obvious and direct ways in which gesture supplements speech were found to be through explicit references to gesture, usually deixis. The other patterns include references to non-deictic gesture, references from non-anaphoric noun phrases and patterns that deviate from the expected one-to-one mapping between gestures and speech. It was observed that gestures or other non verbal cues may enhance the performance of linguistic techniques.
Utilization of multiple channels in ad hoc networks provides the benefits of reducing the probability of collisions, decreasing the normalized propagation delay and support of QoS. However, problems like busy waiting ...
详细信息
Utilization of multiple channels in ad hoc networks provides the benefits of reducing the probability of collisions, decreasing the normalized propagation delay and support of QoS. However, problems like busy waiting or deadlock existence degrade the bandwidth utilization. In this paper, we present the modelling of the transmission-request network as a directed graph in the multi-channel environment. A distributed self-stabilizing algorithm which performs maximal matching is also presented for the new model. By using the algorithm, we propose a new multichannel MAC protocol, named MAXM (MAXimal Matching multi-channel MAC protocol), for ad hoc networks. The idea of MAXM is to maximize utilization of bandwidth by adopting maximal matching. MAXM is deadlock-free, self-stabilizing and able to converge in finite time. Simulation results show that MAXM offers significant improvements over previous work in throughput and transmission delay.
Exception handling is a technology that tests and handles exception. Because of the exception propagation, there are many difficulties to solve in program analysis, understanding, testing and maintenance. Firstly, the...
详细信息
ISBN:
(纸本)0889864276
Exception handling is a technology that tests and handles exception. Because of the exception propagation, there are many difficulties to solve in program analysis, understanding, testing and maintenance. Firstly, the paper proposes a precise and efficient representation, exception propagation graph, for the C++ programs with explicit exception raises and exception handling constructs. The method can easily gain the paths of exception propagation. Then an efficient algorithm is presented to construct the exception propagation graph. The paper also discusses how exception propagation path can be used for structural testing and how insight into exception handling strategies can be gained from it. This provides threads for improving the design and implementation of exception handling codes.
A novel technique for the secret video communicating based on the discrete wavelet transform(DWT) is present. Transform a intra frame of video to three gray image firstly, and then comp.te the Lth-level discrete wavel...
详细信息
ISBN:
(纸本)981238961X
A novel technique for the secret video communicating based on the discrete wavelet transform(DWT) is present. Transform a intra frame of video to three gray image firstly, and then comp.te the Lth-level discrete wavelet decomp.sition of the gray images, with which the secret data W is embed simultaneously and do invert wavelet transform to obtain the gray images which contains the secret information. Change the intra frame of video based the three gray images to make the intra frame contain the secret information. When extract the secret information, transform the intra frame to three gray image and do Lth-level discrete wavelet transform to the gray images, distill the secret data W' from the wavelet coefficients of the three gray images. Test results show the superior performance of the technique and potential for the secret communicating of video.
In this paper, we present a practical UMTS-WLAN interworking architecture based on 3GPP standards, and propose a seamless handoff scheme that guarantees low delay and low packet loss during UMTS-WLAN handoff. For low ...
详细信息
In this paper, we explore new methods by which speakers can be identified and discriminated, using features derived from the fourier transform phase. The Modified Group Delay Feature(MODGDF) which is a parameterized f...
详细信息
In this paper, we explore new methods by which speakers can be identified and discriminated, using features derived from the fourier transform phase. The Modified Group Delay Feature(MODGDF) which is a parameterized form of the modified group delay function is used as a front end feature in this study. A Gaussian mixture model(GMM) based speaker identification system is built with the MODGDF as the front end feature. The system is tested on both clean (TIMIT) and noisy telephone(NTIMIT) speech. The results obtained are comp.red with traditional Mel frequency cepstral coefficients(MFCC) which is derived from the fourier transform magnitude. When both MFCC and MODGDF were combined, the performance improved by about 4% indicating that both phase and magnitude contain comp.ementary information. In an earlier paper, it was shown that the MODGDF does possess phoneme specific characteristics. In this paper we show that the MODGDF has speaker specific properties. We also make an attempt to understand speaker discriminating characteristics of the MODGDF using the nonlinear mapping technique based on Sammon mapping and find that the MODGDF empirically demonstrates a certain level of linear separability among speakers.
This paper introduces a comp.ler-orchestrated prefetching system as a unified framework geared toward ameliorating the gap between processing speeds and memory access latencies. We focus the scope of the optimization ...
详细信息
ISBN:
(纸本)1581138040
This paper introduces a comp.ler-orchestrated prefetching system as a unified framework geared toward ameliorating the gap between processing speeds and memory access latencies. We focus the scope of the optimization on specific subsets of the program dependence graph that succinctly characterize the memory access pattern of both regular array-based applications and irregular pointer-intensive programs. We illustrate how program embedded precomp.tation via speculative execution can accurately predict and effectively prefetch future memory references with negligible overhead. The proposed techniques reduce the total running time of seven SPEC benchmarks and two OLDEN benchmarks by 27% on an Itanium 2 processor. The improvements are in addition to several state-of-the-art optimizations including software pipelining and data prefetching. In addition, we use cycle-accurate simulations to identify important and lightweight architectural innovations that further mitigate the memory system bottleneck. In particular, we focus on the notoriously challenging class of pointer-chasing applications, and demonstrate how they may benefit from a novel scheme of sentineled prefetching. Our results for twelve SPEC benchmarks demonstrate that 45% of the processor stalls that are caused by the memory system are avoidable. The techniques in this paper can effectively mask long memory latencies with little inst.uction overhead, and can readily contribute to the performance of processors today.
暂无评论