Journal updates for 2017.04.30 Biophysical Journal Vol. 112, no. 5 Nothing of interest Vol. 112, no. 6 Deconvolution of Camera Instrument Response Functions John H. Lewis, Ryan M. Jamiolkowski, Michael S. Woody, E. Michael Ostap, Yale E. Goldman Dept. of Physiology, Perelman School of Medicine, University of Pennsylvania Temporal sequences of fluorescence intensities in single-molecule experiments are often obtained from stacks of camera images. The dwell times of different macromolecular structural or functional states, correlated with characteristic fluorescence intensities, are extracted from the images and combined into dwell time distributions that are fitted by kinetic functions to extract corresponding rate constants. The frame rate of the camera limits the time resolution of the experiment and thus the fastest rate processes that can be reliably detected and quantified. However, including the influence of discrete sampling (framing) on the detected time series in the fitted model enables rate processes near to the frame rate to be reliably estimated. This influence, similar to the instrument response function in other types of instruments, such as pulsed emission decay fluorometers, is easily incorporated into the fitted model. The same concept applies to any temporal data that is low-pass filtered or decimated to improve signal to noise ratio. Vol. 112, no. 7 Three-Dimensional Super-Resolution in Eukaryotic Cells Using the Double-Helix Point Spread Function Alexander R. Carr, Aleks Ponjavic, Srinjan Basu, James McColl, Ana Mafalda Santos, Simon Davis, Ernest D. Laue, David Klenerman, Steven F. Lee Depts. of Chemistry and Biochemistry, University of Cambridge, Radcliffe Dept. of Clinical Medicine, University of Oxford, Single-molecule localization microscopy, typically based on total internal reflection illumination, has taken our understanding of protein organization and dynamics in cells beyond the diffraction limit. However, biological systems exist in a complicated three-dimensional environment, which has required the development of new techniques, including the double-helix point spread function (DHPSF), to accurately visualize biological processes. The application of the DHPSF approach has so far been limited to the study of relatively small prokaryotic cells. By matching the refractive index of the objective lens immersion liquid to that of the sample media, we demonstrate DHPSF imaging of up to 15-µm-thick whole eukaryotic cell volumes in three to five imaging planes. We illustrate the capabilities of the DHPSF by exploring large-scale membrane reorganization in human T cells after receptor triggering, and by using single-particle tracking to image several mammalian proteins, including membrane, cytoplasmic, and nuclear proteins in T cells and embryonic stem cells. Vol. 112, no. 8 Nothing of interest. %============================================================= Proceedings of the National Academy of Sciences, USA Vol. 114, no. 14,15,16,17,18 Nothing of interest. %=============================================================
Review of Scientific Instruments Vol. 88, no 4 Edge-preserving reconstruction from sparse projections of limited-angle computed tomography using l0-regularized gradient prior Wei Yu 1, Chengxiang Wang 2, and Min Huang 3,4 1 School of Biomedical Engineering, Hubei University of Science and Technology, Xianning 437100, China 2 School of Mathematical Sciences, University of Electronic Science and Technology of China, Chengdu 611731, China 3 School of Biomedical Engineering, South-Central University for Nationalities, Wuhan 430074, China 4 Hubei Key Laboratory of Medical Information Analysis and Tumor Diagnosis and Treatment, Wuhan 430074, China Accurate images reconstructed from limited computed tomography (CT) data are desired when reducing the X-ray radiation exposure imposed on patients. The total variation (TV), known as the l 1 - norm of the image gradient magnitudes, is popular in CT reconstruction from incomplete projection data. However, as the projection data collected are from a sparse-view of the limited scanning angular range, the results reconstructed by a TV-based method suffer from blocky artifact and gradual changed artifacts near the edges, which in turn make the reconstruction images degraded. Different from the TV, the l0l0-norm of an image gradient counts the number of its non-zero coefficients of the image gradient. Since the regularization based on the l0l0-norm of the image gradient will not penalize the large gradient magnitudes, the edge can be effectively retained. In this work, an edgepreserving image reconstruction method based on l 0 -regularized gradient prior was investigated for limited-angle computed tomography from sparse projections. To solve the optimization model effectively, the variable splitting and the alternating direction method (ADM) were utilized. Experiments demonstrated that the ADM-like method used for the nonconvex optimization problem has better performance than other classical iterative reconstruction algorithms in terms of edge preservation and artifact reduction. Note: An improved low-frequency correction technique for piezoelectric force sensors in high-speed nanopositioning systems Yuen K. Yong a) and Andrew J. Fleming b) School of Electrical Engineering and Computing, University of Newcastle, Callaghan, NSW, Australia Piezoelectric force and position sensors provide high sensitivity but are limited at low frequencies due to their high-pass response which complicates the direct application of integral control. To overcome this issue, an additional sensor or low-frequency correction method is typically employed. However, these approaches introduce an additional first-order response that must be higher than the high-pass response of the piezo and interface electronics. This article describes a simplified method for lowfrequency correction that uses the piezoelectric sensor as an electrical component in a filter circuit. The resulting response is first-order, rather than second-order, with a cut-off frequency equal to that of a buffer circuit with the same input resistance. The proposed method is demonstrated to allow simultaneous damping and tracking control of a high-speed vertical nanopositioning stage. Vol. 88, no. 5 Nothing of interest %=============================================================
Physical Review E Vol. 95, no. 4 Nothing of interest %============================================================= + one: Proc. IEEE International Conference on Electronics, Circuits, and Systems, pp. 313-316, 2016. Yosra Gargouri, Herve Pitt, and Patrick Loumeau Telecom ParisTech In order to reduce power consumption and limit the amount of data acquired and stored for astrophysical signals, an emerging sampling paradigm called compressed sensing (also known as compressive sensing, compressive sampling, CS) could potentially be an efficient solution. The design of radio receiver architecture based on CS requires knowledge of the sparsity domain of the signal and an appropriate measurement matrix. In this paper, we analyze an astrophysical signal (jovian signal with a bandwidth of 40 MHz) by extracting its relevant information via the Radon Transform. Then, we study its sparsity and we establish its sensing modality as well as the minimum number of measurements required. Experimental results demonstrate that our signal is sparse in the frequency domain with a compressibility level of at least 10%. Using the Non Uniform Sampler (NUS) as receiver architecture, we prove that by taking 1/3 of samples at random we can recover the relevant information.
IEEE Transactions on Automatic Control: Vol. 62, Issue 4 Mostly not seemingly relevant IEEE Transactions on Image Processing: Vol. 26, Issue 4 Graph-Based Transform for 2D Piecewise Smooth Signals With Random Discontinuity Locations Dong Zhang ; Jie Liang The graph-based block transform recently emerged as an effective tool for compressing some special signals such as depth images in 3D videos. However, in existing methods, overheads are required to describe the graph of the block, from which the decoder has to calculate the transform via timeconsuming eigendecomposition. To address these problems, in this paper, we aim to develop a single graph-based transform for a class of 2D piecewise smooth signals with similar discontinuity patterns. We first consider the deterministic case with a known discontinuity location in each row. We propose a 2D first-order autoregression (2D AR1) model and a 2D graph for this type of signals. We show that the closed-form expression of the inverse of a biased Laplacian matrix of the proposed 2D graph is exactly the covariance matrix of the proposed 2D AR1 model. Therefore, the optimal transform for the signal are the eigenvectors of the proposed graph Laplacian. Next, we show that similar results hold in the random case, where the locations of the discontinuities in different rows are randomly distributed within a confined region, and we derive the closed-form expression of the corresponding optimal 2D graph Laplacian. The theory developed in this paper can be used to design both pre-computed transforms and signaldependent transforms with low complexities. Finally, depth image coding experiments demonstrate that our methods can achieve similar performance to the state-of-the-art method, but our complexity is much lower. Steerable Wavelet Machines (SWM): Learning Moving Frames for Texture Classification (EPFL) Adrien Depeursinge ; Zsuzsanna Püspöki ; John Paul Ward ; Michael Unser We present texture operators encoding class-specific local organizations of image directions (LOIDs) in a rotation-invariant fashion. The LOIDs are key for visual understanding, and are at the origin of the success of the popular approaches, such as local binary patterns (LBPs) and the scale-invariant feature transform (SIFT). Whereas, LBPs and SIFT yield hand-crafted image representations, we propose to learn data-specific representations of the LOIDs in a rotation-invariant fashion. The image operators are based on steerable circular harmonic wavelets (CHWs), offering a rich and yet compact initial representation for characterizing natural textures. The joint location and orientation required to encode the LOIDs is preserved by using moving frames (MFs) texture representations built from locally-steered image gradients that are invariant to rigid motions. In a second step, we use support vector machines to learn a multi-class shaping matrix for the initial CHW representation, yielding data-driven MFs called steerable wavelet machines (SWMs). The SWM forward function is composed of linear operations (i.e.,
convolution and weighted combinations) interleaved with non-linear steermax operations. We experimentally demonstrate the effectiveness of the proposed operators for classifying natural textures. Our scheme outperforms recent approaches on several test suites of the Outex and the CUReT databases. Graph Laplacian Regularization for Image Denoising: Analysis in the Continuous Domain (Hong Kong Univ. of Sci.&Tech.) Jiahao Pang ; Gene Cheung Inverse imaging problems are inherently underdetermined, and hence, it is important to employ appropriate image priors for regularization. One recent popular prior-the graph Laplacian regularizer-assumes that the target pixel patch is smooth with respect to an appropriately chosen graph. However, the mechanisms and implications of imposing the graph Laplacian regularizer on the original inverse problem are not well understood. To address this problem, in this paper, we interpret neighborhood graphs of pixel patches as discrete counterparts of Riemannian manifolds and perform analysis in the continuous domain, providing insights into several fundamental aspects of graph Laplacian regularization for image denoising. Specifically, we first show the convergence of the graph Laplacian regularizer to a continuous-domain functional, integrating a norm measured in a locally adaptive metric space. Focusing on image denoising, we derive an optimal metric space assuming non-local self-similarity of pixel patches, leading to an optimal graph Laplacian regularizer for denoising in the discrete domain. We then interpret graph Laplacian regularization as an anisotropic diffusion scheme to explain its behavior during iterations, e.g., its tendency to promote piecewise smooth signals under certain settings. To verify our analysis, an iterative image denoising algorithm is developed. Experimental results show that our algorithm performs competitively with state-of-the-art denoising methods, such as BM3D for natural images, and outperforms them significantly for piecewise smooth images. IEEE Transactions on Automation Science and Engineering: Vol. 14, Issue 2 (Apr) Mostly not seemingly relevant
Mechatronics Vol. 44 (June 2017 -- In Progress) Vibration isolator carrying atomic force microscope s head Shingo Ito, Severin Unger, Georg Schitter Christian Doppler Laboratory for Precision Engineering for Automated In-Line Metrology, Automation and Control Institute (ACIN), TU Wien, Gusshausstrasse 27-29, Vienna 1040, Austria For high-resolution imaging in harsh environments this paper proposes a vibration isolation system that actively positions the head of an atomic force microscope (AFM) to maintain the vertical distance to the sample. On the moving platform carrying the AFM head, a displacement sensor is installed to detect the vibrations between the probe and the sample that impair the imaging quality. The detected vibrations are rejected by vertically moving the platform with feedback control. For the motion, flexure-guided Lorentz actuators are designed, such that the resulting suspension mode occurs around the major spectrum of the floor vibrations. By feedback control design, the high gain of the suspension mode is used to increase the open-loop gain for better vibration rejection. The experimental results demonstrate that the vibration isolation system can reject 99.3% of the vibrations. As a result, AFM imaging of nanoscale features is successfully performed in a vibrational environment. +One: Journal of Process Control Vol. 54 (June 2017) Data-driven sensor fault diagnosis systems for linear feedback control loops Kai Wang a, Junghui Chen b,, Zhihuan Song a a State Key Laboratory of Industrial Control Technology, Zhejiang University, Hangzhou, 310027, Zhejiang, China b Department of Chemical Engineering, Chung Yuan Christian University, Chungli, Taoyuan, 32023, Taiwan, ROC This paper develops a sensor fault diagnosis (SFD) scheme for a multi-input and multi-output linear dynamic system under feedback control to identify different types of sensor faults (bias,drift and precision degradation), particularly for the incipient sensor faults. Feedback control, leading to fault propagation and
disguised fault rectification, imposes the challenge on the data-driven SFD. With only available output data in closed loop, the proposed scheme comprises two stages of residual generation and residual evaluation. In the residual generation, a data-driven identification of the residual generator for the feedback control system is proposed. One class of parameters in the residual generator are estimated using process delays while another class of parameters describing the output dynamic are derived by the Bayes formula. The means and variances control charts of online calculated residuals are made to judge the root cause. Two case studies are performed to illustrate the effectiveness of the proposed method.
Journal Updates: May 2017 Automatica Output regulation by error dynamic feedback in hybrid systems with periodic state jumps Volume 81, July 2017, Pages 322 334 Elena Zattoni, Anna Maria Perdon, Giuseppe Conte Department of Electrical, Electronic, and Information Engineering G. Marconi Alma Mater Studiorum University of Bologna, 40136 Bologna, Italy Department of Information Engineering, Polytechnic University of Marche, 60131 Ancona, Italy This work deals with output regulation in multivariable hybrid systems featuring a continuous-time linear dynamics periodically affected by instantaneous changes of the state. More precisely, given a hybrid linear plant and a hybrid linear exogenous system, with periodic state jumps, the problem consists in finding a hybrid feedback regulator, with the same characteristics, achieving global asymptotic stability of the closed-loop dynamics and asymptotic tracking of the reference generated by the exogenous system for all the initial states. Starting from a general, necessary and sufficient condition for the existence of a solution, the discussion leads to a more specific, sufficient condition which outlines the computational framework for a straightforward synthesis of the compensator. The internal model principle is shown to hold in a more general formulation than the original one, adapted to the hybrid systems considered. A numerical example is worked out with the aim of illustrating how to implement the devised technique. The geometric approach is the key methodology in attaining these results. (Brief paper) Observability and diagnosability of finite state systems: A unifying framework Volume 81, July 2017, Pages 115 122 Elena De Santis, Maria Domenica Di Benedetto University of L Aquila, DISIM, Department of Information Engineering, Computer Science and Mathematics, Center of Excellence DEWS, 67100 L Aquila, Italy In this paper, a general framework is proposed for the analysis and characterization of observability and diagnosability of finite state systems. Observability corresponds to the reconstruction of the system s discrete state, while diagnosability corresponds to the possibility of determining the past occurrence of some particular states, for example faulty states. A unifying framework is proposed where observability and diagnosability properties are defined with respect to a critical set, i.e. a set of discrete states representing a set of faults, or more generally a set of interest. These properties are characterized and the involved conditions provide an estimation of the delay required for the detection of a critical state, of the precision of the delay estimation and of the duration of a possible initial transient where the diagnosis is not possible or not required. Our framework makes it possible to precisely compare some of the observability and diagnosability notions existing in the literature with the ones introduced in our paper, and this comparison is presented.
(Brief paper) Event-triggered intermittent sampling for nonlinear model predictive control Volume 81, July 2017, Pages 148 155 Kazumune Hashimoto, Shuichi Adachi, Dimos V. Dimarogonas School of Applied Physics and Physico Informatics, Keio University, Yokohama, Japan School of Electrical Engineering, KTH Royal Institute of Technology, 10044 Stockholm, Sweden In this paper, we propose a new aperiodic formulation of model predictive control for nonlinear continuous-time systems. Unlike earlier approaches, we provide event-triggered conditions without using the optimal cost as a Lyapunov function candidate. Instead, we evaluate the time interval when the optimal state trajectory enters a local set around the origin. The obtained event-triggered strategy is more suitable for practical applications than the earlier approaches in two directions. First, it does not include parameters (e.g., Lipschitz constant parameters of stage and terminal costs) which may be a potential source of conservativeness for the event-triggered conditions. Second, the event-triggered conditions are necessary to be checked only at certain sampling time instants, instead of continuously. This leads to the alleviation of the sensing cost and becomes more suitable for practical implementations under a digital platform. The proposed event-triggered scheme is also validated through numerical simulations. (Brief paper) Optimal control for pointwise asymptotic stability in a hybrid control system Volume 81, July 2017, Pages 397 402 Rafal Goebel Department of Mathematics and Statistics, Loyola University Chicago, 1032 W. Sheridan Road, Chicago, IL 60660, United States Pointwise asymptotic stability, or semistability, is a property of the set of equilibria of a dynamical system, where every equilibrium is Lyapunov stable and every solution is convergent to some equilibrium. Under an appropriate version of asymptotic controllability assumption, it is shown that the property can be achieved in a hybrid control system by openloop optimal solutions of an infinite-horizon optimal control problem. For discrete-time systems, the optimal solutions can be generated by feedback. Regularity of the optimal value function and the existence of hybrid optimal controls are also studied. (Brief paper) Stabilization by using artificial delays: An LMI approach Volume 81, July 2017, Pages 429 437 Emilia Fridman, Leonid Shaikhet School of Electrical Engineering, Tel-Aviv University, Tel-Aviv 69978, Israel Static output-feedback stabilization for the nth order vector differential equations by using artificial multiple delays is considered. Under assumption of the stabilizability of the system by a static feedback that depends on the output and its derivatives up to the order n 1, a delayed static output-feedback is found that stabilizes the system. The conditions for the stability analysis of the resulting closed-loop system are given in terms of simple LMIs. It is shown that the LMIs are always feasible for appropriately chosen gains and small enough delays. Robust stability analysis in the presence of uncertain time-varying delays and stochastic perturbation of the system coefficients is provided. Numerical examples including chains of three and four integrators that are stabilized by static output-feedbacks with multiple delays illustrate the efficiency of the method.
System & Control Letters (Highlighted Paper)Autocovariance-based plant-model mismatch estimation for linear model predictive control Volume 104, June 2017, Pages 5 14 Siyun Wang, Jodie M. Simkoff, Michael Baldea, Leo H. Chiang, Ivan Castillo, Rahul Bindlish, David B. Stanley McKetta Department of Chemical Engineering, The University of Texas at Austin, Austin, TX 78712, USA The Dow Chemical Company, Freeport, TX 77541, USA In this paper, we present autocovariance-based estimation as a novel methodology for determining plant-model mismatch for multiple-input, multiple-output systems operating under model predictive control. Considering discretetime, linear time invariant systems under reasonable assumptions, we derive explicit expressions of the autocovariances of the system inputs and outputs as functions of the plant-model mismatch. We then formulate the mismatch estimation problem as a global optimization aimed at minimizing the discrepancy between the theoretical autocovariance estimates and the corresponding values computed from historical closed-loop operating data. Practical considerations related to implementing these ideas are discussed, and the results are illustrated with a chemical process case study. Nonlinear sliding mode control design: An LMI approach Volume 104, June 2017, Pages 38 44 Alán Tapia, Miguel Bernal, Leonid Fridman National Autonomous University of Mexico, Department of Control Engineering and Robotics, Division of Electrical Engineering, Engineering Faculty, C.P. 04510, Mexico City, Mexico Sonora Institute of Technology, 5 de Febrero 818 Sur, C.P. 85000, Cd. Obregon, Sonora, Mexico This paper presents a novel nonlinear sliding mode control methodology for systems with both matched and unmatched perturbations (including parametric uncertainties). Instead of traditional approaches where uncertainties and nonlinearities are coped with via linear nominal models and linear sliding surfaces, the proposed approach incorporates exact convex expressions to represent both the nonlinear surface and the system, thus allowing a significant chattering reduction. Moreover, thanks to the convex form of the nonlinear nominal model, when combined with the direct Lyapunov method, it leads to linear matrix inequalities, which are efficiently solved via convex optimization techniques. Illustrative examples are provided.
IEEE Transactions on Robotics A Switched Systems Framework for Guaranteed Convergence of Image-Based Observers With Intermittent Measurements Volume: 33, Issue: 2, April 2017, Page(s): 266-280 Anup Parikh ; Teng-Hu Cheng ; Hsi-Yuan Chen ; Warren E. Dixon Department of Mechanical and Aerospace Engineering, University of Florida, Gainesville, FL, USA : Switched systems theory is used to analyze the stability of image-based observers for three-dimensional localization of objects in a scene in the presence of intermittent measurements due to occlusions, feature tracking losses, or a limited camera field of view, for example. Generally, observers or filters that are exponentially stable under persistent measurement availability may have unbounded error growth under intermittent measurement loss, even while providing seemingly accurate state estimates. By constructing a framework that utilizes a state predictor during periods when measurements are not available, a class of image-based observers is shown to be exponentially convergent in the presence of intermittent measurements if an average dwell time, and a total unmeasurability time, condition is satisfied. The conditions are developed in a general form, applicable to any observer that is exponentially convergent assuming persistent visibility, and utilizes object motion knowledge to reduce the amount of time measurements must be available to maintain convergence guarantees. Based on the stability results, simulations are provided to show improved performance compared to a zero-order hold approach, where state estimates are held constant when measurements are not available. Experimental results are also included to verify the theoretical results, to demonstrate applicability of the developed observer and predictor design, and to compare against a typical approach using an extended Kalman filter. The Impact of Diversity on Optimal Control Policies for Heterogeneous Robot Swarms Volume: 33, Issue: 2, April 2017, Page(s): 346-358 Amanda Prorok ; M. Ani Hsieh ; Vijay Kumar University of Pennsylvania, Philadelphia, PA, USA : We consider the problem of distributing a large group of heterogeneous robots among a set of tasks that require specialized capabilities in order to be completed. We model the system of heterogeneous robots as a community of species, in which each species (robot type) is defined by the traits (capabilities) that it owns. In order to solve the distribution problem, we develop centralized as well as decentralized methods to efficiently control the heterogeneous swarm of robots. Our methods assume knowledge of the underlying task topology and are based on a continuous model of the system that defines transition rates to and from tasks, for each robot species. Our optimization of the transition rates is fully scalable with respect to the number of robots, number of species, and number of traits. Building on this result, we propose a real-time optimization method that enables an online adaptation of transition rates as a function of the state of the current robot distribution. We also show how the robot distribution can be approximated based on local information only, consequently enabling the development of a decentralized controller. We evaluate our methods by means of microscopic simulations and show how the performance of the latter is well predicted by the macroscopic equations. Importantly, our framework also includes a diversity metric that enables an evaluation of the impact of swarm heterogeneity on performance. The metric defines the notion of minspecies, i.e., the minimum set of species that are required to achieve a given goal. We show that two distinct goal functions lead to two specializations of minspecies, which we term as eigenspecies and coverspecies. Quantitative results show the relation between diversity and performance.
Mobile Robot Path Planners With Memory for Mobility Diversity Algorithms Volume: 33, Issue: 2, April 2017, Page(s): 419-431 Daniel Bonilla Licea ; Des McLernon ; Mounir Ghogho University of Leeds, Leeds, U.K. : Mobile robots (MRs) using wireless communications often experience small-scale fading so that the wireless channel gain can be low. If the channel gain is poor (due to fading), the robot can move (a small distance) to another location to improve the channel gain and so compensate for fading. Techniques using this principle are called mobility diversity algorithms (MDAs). MDAs intelligently explore a number of points to find a location with high channel gain while using little mechanical energy during the exploration. Until now, the location of these points has been predetermined. In this paper, we show how we can adapt their positions by using channel predictors. Our results show that MDAs, which adapt the location of those points, can in fact outperform (in terms of the channel gain obtained and mechanical energy used) the MDAs that use predetermined locations for those points. These results will significantly improve the performance of the MDAs and consequently allow MRs to mitigate poor wireless channel conditions in an energy-efficient manner. Multiobjective Optimization Based on Expensive Robotic Experiments under Heteroscedastic Noise Volume: 33, Issue: 2, April 2017, Page(s): 468-483 Ryo Ariizumi ; Matthew Tesch ; Kenta Kato ; Howie Choset ; Fumitoshi Matsuno Department of Mechanical Science and Engineering, Graduate School of Engineering, Nagoya University, Nagoya, Japan Robotics Institute, Carnegie Mellon University, Pittsburgh, PA, USA Department of Mechanical Engineering and Science, Graduate School of Engineering, Kyoto University, Kyoto, Japan : In many engineering problems, including those related to robotics, optimization of the control policy for multiple conflicting criteria is required. However, this can be very challenging because of the existence of noise, which may be input dependent or heteroscedastic, and restrictions regarding the number of evaluations owing to the costliness of the experiments in terms of time and/or money. This paper presents a multiobjective optimization algorithm for expensiveto-evaluate noisy functions for robotics. We present a method for model selection between heteroscedastic and standard homoscedastic Gaussian process regression techniques to create suitable surrogate functions from noisy samples, and to find the point to be observed at the next step. This algorithm is compared against an existing multiobjective optimization algorithm, and then used to optimize the speed and head stability of the sidewinding gait of a snake robot.
IEEE Transactions on Signal Processing (Issue 11, 12, 13, 2017) IEEE Transactions on Image Processing (Issue 4, 5, 6, 2017) L 0 Gradient Projection Shunsuke Ono, Member, IEEE the Laboratory for Future Interdisciplinary Research of Science and Technology, Institute of Innovative Research, Tokyo Institute of Technology, Kanagawa 226-8503, Japan Minimizing L 0 gradient, the number of the non-zero gradients of an image, together with a quadratic datafidelity to an input image has been recognized as a powerful edge-preserving filtering method. However, the L 0 gradient minimization has an inherent difficulty: a user-given parameter controlling the degree of flatness does not have a physical meaning since the parameter just balances the relative importance of the L 0 gradient term to the quadratic data-fidelity term. As a result, the setting of the parameter is a troublesome work in the L 0 gradient minimization. To circumvent the difficulty, we propose a new edge-preserving filtering method with a novel use of the L 0 gradient. Our method is formulated as the minimization of the quadratic data-fidelity subject to the hard constraint that the L 0 gradient is less than a user-given parameter α. This strategy is much more intuitive than the L 0 gradient minimization because the parameter has a clear meaning: the L 0 gradient value of the output image itself, so that one can directly impose a desired degree of flatness by α. We also provide an efficient algorithm based on the so-called alternating direction method of multipliers for computing an approximate solution of the nonconvex problem, where we decompose it into two subproblems and derive closed-form solutions to them. The advantages of our method are demonstrated through extensive experiments. Low-Rank Embedding for Robust Image Feature Extraction Wai Keung Wong, Zhihui Lai, Jiajun Wen, Xiaozhao Fang, and Yuwu Lu the Institute of Textiles and Clothing, The Hong Kong Polytechnic University, Hong Kong Robustness to noises, outliers, and corruptions is an important issue in linear dimensionality reduction. Since the sample-specific corruptions and outliers exist, the class-special structure or the local geometric structure is destroyed, and thus, many existing methods, including the popular manifold learning-based linear dimensionality methods, fail to achieve good performance in recognition tasks. In this paper, we focus on the unsupervised robust linear dimensionality reduction on corrupted data by introducing the robust low-rank representation (LRR). Thus, a robust linear dimensionality reduction technique termed low-rank embedding (LRE) is proposed in this paper, which provides a robust image representation to uncover the potential relationship among the images to reduce the negative influence from the occlusion and corruption so as to enhance the algorithms robustness in image feature extraction. LRE searches the optimal LRR and optimal subspace simultaneously. The model of LRE can be solved by alternatively iterating the argument Lagrangian multiplier method and the eigendecomposition. The theoretical analysis, including convergence analysis and computational complexity, of the algorithms is presented. Experiments on some well-known databases with different corruptions show that LRE is superior to the previous methods of feature extraction, and therefore, it indicates the robustness of the proposed method. 1
Classification via Sparse Representation of Steerable Wavelet Frames on Grassmann Manifold: Application to Target Recognition in SAR Image Ganggang Dong ; Gangyao Kuang ; Na Wang ; Wei Wang the College of Electronics Science and Engineering, National University of Defense Technology, Changsha 410073, China Automatic target recognition has been widely studied over the years, yet it is still an open problem. The main obstacle consists in extended operating conditions, e.g.., depression angle change, configuration variation, articulation, and occlusion. To deal with them, this paper proposes a new classification strategy. We develop a new representation model via the steerable wavelet frames. The proposed representation model is entirely viewed as an element on Grassmann manifolds. To achieve target classification, we embed Grassmann manifolds into an implicit reproducing Kernel Hilbert space (RKHS), where the kernel sparse learning can be applied. Specifically, the mappings of training sample in RKHS are concatenated to form an overcomplete dictionary. It is then used to encode the counterpart of query as a linear combination of its atoms. By designed Grassmann kernel function, it is capable to obtain the sparse representation, from which the inference can be reached. The novelty of this paper comes from: 1) the development of representation model by the set of directional components of Riesz transform; 2) the quantitative measure of similarity for proposed representation model by Grassmann metric; and 3) the generation of global kernel function by Grassmann kernel. Extensive comparative studies are performed to demonstrate the advantage of proposed strategy. Dynamical Textures Modeling via Joint Video Dictionary Learning Xian Wei ; Yuanxiang Li ; Hao Shen ; Fang Chen ; Martin Kleinsteuber ; Zhongfeng Wang School of Aeronautics and Astronautics, Shanghai Jiao Tong University, Shanghai, China Video representation is an important and challenging task in the computer vision community. In this paper, we consider the problem of modeling and classifying video sequences of dynamic scenes which could be modeled in a dynamic textures (DTs) framework. At first, we assume that image frames of a moving scene can be modeled as a Markov random process. We propose a sparse coding framework, named joint video dictionary learning (JVDL), to model a video adaptively. By treating the sparse coefficients of image frames over a learned dictionary as the underlying states, we learn an efficient and robust linear transition matrix between two adjacent frames of sparse events in time series. Hence, a dynamic scene sequence is represented by an appropriate transition matrix associated with a dictionary. In order to ensure the stability of JVDL, we impose several constraints on such transition matrix and dictionary. The developed framework is able to capture the dynamics of a moving scene by exploring both the sparse properties and the temporal correlations of consecutive video frames. Moreover, such learned JVDL parameters can be used for various DT applications, such as DT synthesis and recognition. Experimental results demonstrate the strong competitiveness of the proposed JVDL approach in comparison with the state-of-the-art video representation methods. Especially, it performs significantly better in dealing with DT synthesis and recognition on heavily corrupted data. 2