Search

This paper is concerned with the quantized identification of rational systems, where the systems’ output is quantized by a logarithmic quantizer. Under the assumption that the systems’ input is periodic, the identification procedure is categorized into two steps. The first step is to identify the noise-free output of systems based on the quantized data. The second is to identify the unknown parameter based on the input and the estimation of the noise-free output. The identification algorithm is also summarized. Asymptotic convergence of the estimators is analyzed in detail, which shows that the estimators are convergent almost everywhere. A numerical example is given to illustrate the results obtained in this paper.

This paper considers the observer design problem for a class of discrete-time system whose nonlinear time-varying terms satisfy incremental quadratic constraints. We first construct a circle criterion based full-order observer by injecting output estimation error into the observer nonlinear terms. We also construct a reduced-order observer to estimate the unmeasured system state. The proposed observers guarantee exponential convergence of the state estimation error to zero. The design of the proposed observers is reduced to solving a set of linear matrix inequalities. It is proved that the conditions under which a full-order observer exists also guarantee the existence of a reduced-order observer. Compared to some previous results in the literature, this work considers a larger class of nonlinearities and unifies some related observer designs for discrete-time nonlinear systems. Finally, a numerical example is included to illustrate the effectiveness of the proposed design.

In this paper, we mainly focus on the problem of quantized feedback stabilization of a discrete-time linear system with Markovian jump packet losses. Based on the packet loss dependent Lyapunov functions, the coarsest quantizer is analyzed in detail to guarantee the mean square quadratic stabilization of the system. If the system adopts the logarithmic quantizer, the necessary and sufficient conditions, which make the system mean square quadratically stabilized by a linear state feedback controller, can be converted into an algebraic Riccati equation and then into an LMI. Then based on this LMI, we obtain the infimum quantization density of the logarithmic quantizer over all dependent Lyapunov functions. Finally, a numerical example is presented to illustrate the effectiveness of the results obtained in this paper.

This paper addresses the feedback stabilization problem for Markov jump linear systems with quantized control input. A simple but powerful input controller for eliminating the input quantization in Markov jump linear systems is used. The proposed mode-dependent input controller comprises two parts: the linear part and the nonlinear part, where the linear part determines the fundamental characteristics of the systems, and the nonlinear part eliminates the effect of the input quantization. The stability is analyzed in detail for both the discrete-time and continuous-time Markov jump linear systems with input quantization. The sufficient conditions and the specific mode-dependent input controller to guarantee the stability of the systems are obtained. Numerical examples are given to illustrate the effectiveness of the results.

For the logarithmic quantizer, under the condition that the input of the quantizer is a continuous random variable, some statistical properties of the quantization output, such as probability density function, characteristic function and moments are analyzed in detail. Then if the input contains unknown parameters, the moment estimate of the parameters can be obtained based on the quantization output observations. Since finite logarithmic quantizer is more feasible in practice, the statistical properties of the output for such quantizer are also analyzed. Two examples are given to illustrate the methods proposed in this paper.

In this paper, we mainly focus on the stabilization problem of discrete-time linear systems with both quantized input feedback and arbitrary packet losses. The coarsest quantization strategy is analyzed in detail to guarantee the asymptotic stability of the system. If the coarsest quantizer is logarithmic, the necessary and sufficient conditions to asymptotically stabilize such system are converted into some algebraic Riccati equations and then into some LMIs. Then the infimum of the quantization density for the logarithmic quantizer over all packet loss dependent Lyapunov functions is obtained according to these LMIs. In addition, we also prove that the sector bound approach for the logarithmic quantizer is still valid for the system with arbitrary packet losses. The problem of asymptotic stability can be converted into a robust control problem with sector bound uncertainties. The robust stability of the uncertain system is formulated into some LMIs. Finally, an example is given to show the effectiveness of the results obtained in this paper.

Based on the extracted contours of objects from images, one point to point (P2P) method is proposed for shape matching and image retrieval. Taking contour of one object as reference, contour of another object is transformed to reach their best match, during which the similarity is evaluated by comparison of two corresponding sets of contour points. Translation, scaling and rotation are all considered in transformation, thus the algorithm is robust to objects with different position, size and posture. The experimental results are presented and compared with those from two popular shape based techniques, Hu invariant moments and Zernike moments. Performance of our new approach has proved its efficiency in both ma tching accuracy and computational expense, and it can be used in related applications together with the other kinds of shape features or eve n color features, texture features, etc.

In this paper, we mainly focus on the problem of quantised feedback stabilisation of a stochastic discrete-time linear system with white noise input. The dynamic quantiser is used here. The stability of the system under state quantisation and input quantisation is analysed in detail, respectively. Both the convergence of the state's mean and the boundedness of the state's covariance matrix norm should be considered when analysing its stability. It is shown that for the two situations of the state quantisation and the input quantisation, if the system without noise input can be stabilised by a linear feedback law, it must be stabilised by the dynamic quantised feedback control policy. The sufficient conditions that the dynamic quantiser should satisfy are given. Using the results obtained in this paper, one can test whether the stochastic system is stabilisable or not. Numerical examples are given to show the effectiveness of the results.

In this paper, our main concern is the problem of stabilization of discrete-time linear systems with both quantization and noise input. Stabilizability by means of quantizers that perform adaptations called “zooming” is analyzed. The noise input can be separated into an additive white noise part and a deterministic constant input disturbance. The analysis of mean-square stability of the system with noise input is essentially the asymptotic stability of the system disturbed by a constant input. It is shown that the system with the constant input disturbance is asymptotically stabilized by the quantized feedback control policy if the system without noise input can be stabilized by a linear state feedback law. Both the state quantization and the input quantization are studied.

Human motion denoising is an indispensable step of data preprocessing for many motion data based applications. In this paper, we propose a data-driven based human motion denoising method that sparsely selects the most correlated subset of motion bases for clean motion reconstruction. Meanwhile, it takes the statistic property of two common noises, i.e., Gaussian noise and outliers, into account in deriving the objective functions. In particular, our method firstly divides each human pose into five partitions termed as poselets to gain a much fine-grained pose representation. Then, these poselets are reorganized into multiple overlapped poselet groups using a lagged window moving across the entire motion sequence to preserve the embedded spatial-temporal motion patterns. Afterward, five compacted and representative motion dictionaries are constructed in parallel by means of fast K-SVD in the training phase; they are used to remove the noise and outliers from noisy motion sequences in the testing phase by solving ?1-minimization problems. Extensive experiments show that our method outperforms its competitors. More importantly, compared with other data-driven based method, our method does not need to specifically choose the training data, it can be more easily applied to real-world applications. HighlightsA fine-grained pose representation model is proposed to boost the performance.We present a data-driven based motion denoising method by solving ?1-minimization problems.The proposed model selects the most correlated motion bases for motion reconstruction.Our method does not need to choose the training data and can be implemented in parallel mode.