This paper presents a new superpixel-based hand gesture recognition system based on a novel superpixel earth mover's distance metric, together with Kinect depth camera. The depth and skeleton ...information from Kinect are effectively utilized to produce markerless hand extraction. The hand shapes, corresponding textures and depths are represented in the form of superpixels, which effectively retain the overall shapes and color of the gestures to be recognized. Based on this representation, a novel distance metric, superpixel earth mover's distance (SP-EMD), is proposed to measure the dissimilarity between the hand gestures. This measurement is not only robust to distortion and articulation, but also invariant to scaling, translation and rotation with proper preprocessing. The effectiveness of the proposed distance metric and recognition algorithm are illustrated by extensive experiments with our own gesture dataset as well as two other public datasets. Simulation results show that the proposed system is able to achieve high mean accuracy and fast recognition speed. Its superiority is further demonstrated by comparisons with other conventional techniques and two real-life applications.
Usually, direction-of-arrival (DOA) estimators are derived under the assumption of uniform white noise, whose covariance matrix is a scaled identity matrix. However, in practice, the noise can be ...nonuniform with an arbitrary unknown diagonal covariance matrix. In this situation, the performance of DOA estimators may be deteriorated considerably if the noise nonuniformity is ignored. To tackle this problem, iterative approaches to subspace estimation are developed and the corresponding subspace-based DOA estimators are addressed. In our proposed methods, the signal subspace and noise covariance matrix are first determined by maximizing the log-likelihood (LL) function or solving a least-squares (LS) minimization problem, both of which are accomplished in an iterative manner. Then, the DOAs are determined from the subspace estimate and/or noise covariance matrix estimate with the help of traditional DOA estimators. As the signal subspace and noise covariance matrix can be computed in closed-form in each iteration, the proposals are computationally attractive. Furthermore, the signal subspace is directly calculated without the requirement of the exact knowledge of the array manifold, enabling us to handle array uncertainties by incorporating conventional subspace-based calibration algorithms. Simulations and experimental results are included to demonstrate the superiority of the proposed approaches.
A new method for direction finding with partly calibrated uniform linear arrays (ULAs) is presented. It is based on the conventional estimation of signal parameters via rotational invariance ...techniques (ESPRIT) by modeling the imperfections of the ULAs as gain and phase uncertainties. For a fully calibrated array, it reduces to the conventional ESPRIT algorithm. Moreover, the direction-of-arrivals (DOAs), unknown gains, and phases of the uncalibrated sensors can be estimated in closed form without performing a spectral search. Hence, it is computationally very attractive. The Cramér-Rao bounds (CRBs) of the partly calibrated ULAs are also given. Simulation results show that the root mean squared error (RMSE) performance of the proposed algorithm is better than the conventional methods when the number of uncalibrated sensors is large. It also achieves satisfactory performance even at low signal-to-noise ratios (SNRs).
This paper proposes a compact, high-linearity, and reconfigurable continuous-time filter with a wide frequency-tuning capability for biopotential conditioning. It uses an active filter topology and a ...new operational-transconductance-amplifier (OTA)-based current-steering (CS) integrator. Consequently, a large time constant τ , good linearity, and linear bandwidth tuning could be achieved in the presented filter with a small silicon area. The proposed filter has a reconfigurable structure that can be operated as a low-pass filter (LPF) or a notch filter (NF) for different purposes. Based on the novel topology, the filter can be readily implemented monolithically and a prototype circuit was fabricated in the 0.18 μm standard complementary-metal-oxide-semiconductor (CMOS) process. It occupied a small area of 0.068 mm
and consumed 25 μW from a 1.8 V supply. Measurement results show that the cutoff frequency of the LPF could be linearly tuned from 0.05 Hz to 300 Hz and the total-harmonic-distortion (THD) was less than -76 dB for a 2 Hz, 200 mVpp sine input. The input-referred noises were 5.5 μVrms and 6.4 μVrms for the LPF and NF, respectively. A comparison with conventional designs reveals that the proposed design achieved the lowest harmonic distortion and smallest on-chip capacitor. Moreover, its ultra-low cutoff frequency and relatively linear frequency tuning capability make it an attractive solution as an analog front-end for biopotential acquisitions.
DOA Estimation and Tracking of ULAs with Mutual Coupling Liao, Bin; Zhang, Zhi-Guo; Chan, Shing-Chow
IEEE transactions on aerospace and electronic systems,
2012-Jan., 2012-01-00, 20120101, Volume:
48, Issue:
1
Journal Article
Peer reviewed
A class of subspace-based methods for direction-of-arrival (DOA) estimation and tracking in the case of uniform linear arrays (ULAs) with mutual coupling is proposed. By treating the ...angularly-independent mutual coupling as angularly-dependent complex array gains, the middle subarray is found to have the same complex array gains. Using this property, a new way for parameterizing the steering vector is proposed and the corresponding method for joint estimation of DOAs and mutual coupling matrix (MCM) using the whole array data is derived based on subspace principle. Simulation results show that the proposed algorithm has a better performance than the conventional subarray-based method especially for weak signals. Furthermore, to achieve low computational complexity for online and time-varying DOA estimation, three subspace tracking algorithms with different arithmetic complexities and tracking abilities are developed. More precisely, by introducing a better estimate of the subspace to the conventional tracking algorithms, two modified methods, namely modified projection approximate subspace tracking (PAST) (MPAST) and modified orthonormal PAST (MOPAST), are developed for slowly changing subspace, whereas a Kalman filter with a variable number of measurements (KFVM) method for rapidly changing subspace is introduced. Simulation results demonstrate that these algorithms offer high flexibility and effectiveness for tracking DOAs in the presence of mutual coupling.
We recently proposed an ESPRIT-like method for direction finding with partly calibrated uniform linear arrays. It has been shown that the unknown gains/phases and directionof-arrivals (DOAs) can be ...estimated jointly. However, this approach relies on the assumption of uniform white noise, i.e., all sensor noise powers are identical. Besides, at most M -2 sources can be handled for an M-element array. This motives us to develop enhanced methods for circumventing the limitations above. This paper extends the uniform white noise to nonuniform noise on one hand, and, on the other hand, allows us to estimate up to M - 1 DOAs of uncorrelated signals. More exactly, for uncorrelated signals, the array can be calibrated according to the specific structure of the array covariance matrix, and the nonuniformity of sensor noises can then be simply eliminated by reformulating the calibrated array covariance matrix. For correlated signals, the nonuniformity of sensor noises is mitigated by solving a least squares minimization problem, such that the signal/noise subspace can be properly determined. Thus, our previously developed ESPRIT-like approach can be adopted to determine the DOAs. The effectiveness of the proposed methods is confirmed by numerical examples.
This paper proposes an adaptive fading Bayesian unscented Kalman filter (AF-BUKF) and explores its application for state estimation of unmanned aircraft systems (UASs). In the AF-BUKF, the state and ...noise densities are approximated as finite Gaussian mixtures, in which the mean and covariance for each component are recursively estimated using the UKF. To avoid the prohibitive computational complexity caused by the exponential growth of mixture components, a Gaussian mixture simplification algorithm is employed. Moreover, the AF-BUKF algorithm employs a novel adaptive fading strategy to recursively update the Gaussian components, so that the adverse effect of inexact knowledge of the state and measurement noise covariance can be mitigated. An AF-BUK Smoother (AF-BUKS) is also proposed by extending the AF-BUKF algorithm using the concept of optimal Bayesian smoothing and the Rauch-Tung-Striebel Smoother to improve estimation accuracy. Experimental results on simulated and real UAS data show that the proposed AF-BUKF/S algorithms can achieve better performance compared with the conventional methods. Thus, they can serve as attractive alternative approaches for nonlinear state estimation of UASs and other problems.
Recently, we considered the problem of direction finding with partly calibrated uniform linear arrays (ULAs) with unknown gains and phases and proposed an ESPRIT-like method for direction-of-arrival ...(DOA) estimation. It was shown that the DOAs, together with unknown sensor gains and phases in the uncalibrated portion of the array, can be estimated in closed form. However, the identifiability of DOA estimation has not yet been addressed. Moreover, though the proposed method performs better than existing ones, it uses the overlapping subarrays only. Thus it is possible to further improve the performance if the whole array aperture is employed. To fill this gap, two main issues are addressed in this paper. First, the ESPRIT-like algorithm is reinvestigated and conditions ensuring the uniqueness of DOA estimates and identifiability are derived. Second, by exploiting the subspace principle, a refining scheme is proposed that is able to improve the performance of the ESPRIT-like algorithm. Numerical examples are carried out to demonstrate the identifiability issue and performance of the refinement.
This article proposes a new recursive linearly constrained minimum variance (LCMV) beamformer based on the extended instrumental variable (EIV) method for planar radial coprime arrays (PRCAs) under ...spatially colored noise. The proposed recursive LCMV beamformer is able to deal with multiple constraints with high precision and low complexity and can be applicable to various array geometrical configurations. Taking advantage of the EIV vector, the proposed beamformer can effectively combat the additive color noise with unknown noise covariance matrix. We develop our recursive LCMV beamformer based on the square-root (SR) EIV algorithm due to its improved numerical stability than the conventional EIV-based algorithms. Furthermore, we studied a class of planar arrays called PRCAs, which consists of a set of linear coprime arrays arranged radially at various azimuth angles. The coprime array property is utilized to enlarge the array aperture leading to higher resolution and stronger interference rejection and it offers additional flexibility in the tradeoffs between array complexity and performance. Simulation results demonstrate that the proposed recursive SREIV-based LCMV beamformer outperforms the conventional QR decomposition based LCMV beamformers in the resolution and suppression of interferences under various scenarios. The PRCAs tested outperform the uniform rectangular arrays with the same number of elements. Moreover, better performance can be achieved with more linear subarrays at the expense of increased complexity.
Automatic lung cancer diagnosis from computer tomography (CT) images requires the detection of nodule location as well as nodule malignancy prediction. This article proposes a joint lung nodule ...detection and classification network for simultaneous lung nodule detection, segmentation and classification subject to possible label uncertainty in the training set. It operates in an end-to-end manner and provides detection and classification of nodules simultaneously together with a segmentation of the detected nodules. Both the nodule detection and classification subnetworks of the proposed joint network adopt a 3-D encoder-decoder architecture for better exploration of the 3-D data. Moreover, the classification subnetwork utilizes the features extracted from the detection subnetwork and multiscale nodule-specific features for boosting the classification performance. The former serves as valuable prior information for optimizing the more complicated 3D classification network directly to better distinguish suspicious nodules from other tissues compared with direct backpropagation from the decoder. Experimental results show that this co-training yields better performance on both tasks. The framework is validated on the LUNA16 and LIDC-IDRI datasets and a pseudo-label approach is proposed for addressing the label uncertainty problem due to inconsistent annotations/labels. Experimental results show that the proposed nodule detector outperforms the state-of-the-art algorithms and yields comparable performance as state-of-the-art nodule classification algorithms when classification alone is considered. Since our joint detection/recognition approach can directly detect nodules and classify its malignancy instead of performing the tasks separately, our approach is more practical for automatic cancer and nodules detection.