Display Method: |
Fine-grained image classification, which aims to distinguish images with subtle distinctions, is a challenging task for two main reasons: lack of sufficient training data for every class and difficulty in learning discriminative features for representation. In this paper, to address the two issues, we propose a two-phase framework for recognizing images from unseen fine-grained classes, i.e., zero-shot fine-grained classification. In the first feature learning phase, we finetune deep convolutional neural networks using hierarchical semantic structure among fine-grained classes to extract discriminative deep visual features. Meanwhile, a domain adaptation structure is induced into deep convolutional neural networks to avoid domain shift from training data to test data. In the second label inference phase, a semantic directed graph is constructed over attributes of fine-grained classes. Based on this graph, we develop a label propagation algorithm to infer the labels of images in the unseen classes. Experimental results on two benchmark datasets demonstrate that our model outperforms the state-of-the-art zero-shot learning models. In addition, the features obtained by our feature learning model also yield significant gains when they are used by other zero-shot learning models, which shows the flexility of our model in zero-shot fine-grained classification.
Image registration is an indispensable component in multi-source remote sensing image processing. In this paper, we put forward a remote sensing image registration method by including an improved multi-scale and multi-direction Harris algorithm and a novel compound feature. Multi-scale circle Gaussian combined invariant moments and multi-direction gray level co-occurrence matrix are extracted as features for image matching. The proposed algorithm is evaluated on numerous multi-source remote sensor images with noise and illumination changes. Extensive experimental studies prove that our proposed method is capable of receiving stable and even distribution of key points as well as obtaining robust and accurate correspondence matches. It is a promising scheme in multi-source remote sensing image registration.
In this paper, the auto-tuning of a fractional order proportional and integral (FOPI) controller is proposed and experimentally validated for two-input two-output (TITO) processes. The proposed method first identifies an unknown TITO plant into fractional order TITO model with time delay. Furthermore, decoupling the TITO process into two fractional order single-input single-output (SISO) transfer function models makes it easier for designing the decentralized FOPI controllers. The proposed control method is a generalization of both integer order and fractional order TITO systems depending on the value of the order of the model. One advantage of this method is the non-requirement of a-priori information of gain and phase crossover frequencies of the system while tuning the controllers. The proposed algorithm is validated both by simulation of a class of TITO process models as well as by experimental analysis of a coupled tank system (CTS).
In order to improve the accuracy of the data fusion filter, a tightly-coupled ultra wide band (UWB)/inertial navigation system (INS)-integrated scheme for indoor human navigation will be investigated in this paper. In this scheme, the data fusion filter employs the difference between the INS-measured and UWB-measured distances as the observation. Moreover, the predictive adaptive Kalman filter (PAKF) for the tightly-coupled INS/UWB-integrated human tracking model with missing data of the UWB-measured distance will be designed, which considers the missing data of the UWB-based distance and employs the predictive UWB-measured distance. Real test results will be done to compare the performance of the Kalman filter (KF), adaptive Kalman filter (AKF), and the PAKF. The test results show that the performance of the AKF is better than the KF. Moreover, the proposed PAKF is able to maintain the performance of the filter when the UWB-based measurement is unavailable.
In this paper, a new adaptive hierarchical sliding mode control scheme for a 3D overhead crane system is proposed. A controller is first designed by the use of a hierarchical structure of two first-order sliding surfaces represented by two actuated and un-actuated subsystems in the bridge crane. Parameters of the controller are then intelligently estimated, where uncertain parameters due to disturbances in the 3D overhead crane dynamic model are proposed to be represented by radial basis function networks whose weights are derived from a Lyapunov function. The proposed approach allows the crane system to be robust under uncertainty conditions in which some uncertain and unknown parameters are highly difficult to determine. Moreover, stability of the sliding surfaces is proved to be guaranteed. Effectiveness of the proposed approach is then demonstrated by implementing the algorithm in both synthetic and real-life systems, where the results obtained by our method are highly promising.
This paper investigates the necessity of feasibility considerations in a fault tolerant control system using the constrained control allocation methodology where both static and dynamic actuator constraints are considered. In the proposed feasible control allocation scheme, the constrained model predictive control (MPC) is employed as the main controller. This considers the admissible region of the control allocation problem as its constraints. Using the feasibility notion in the control allocation problem provides the main controller with information regarding the actuator′s status, which leads to closed loop system performance improvement. Several simulation examples under normal and faulty conditions are employed to illustrate the effectiveness of the proposed methodology. The main results clearly indicate that closed loop performance and stability characteristics can be significantly degraded by neglecting the actuator constraints in the main controller. Also, it is shown that the proposed strategy substantially enlarges the domain of attraction of the MPC combined with the control allocation as compared to the conventional MPC.
The objective of this paper is to propose a reduced-order observer for a class of Lipschitz nonlinear discrete-time systems. The conditions that guarantee the existence of this observer are presented in the form of linear matrix inequalities (LMIs). To handle the Lipschitz nonlinearities, the Lipschitz condition and the Young′s relation are adequately operated to add more degrees of freedom to the proposed LMI. Necessary and sufficient conditions for the existence of the unbiased reduced-order observer are given. An extension to
Manufacturing features represent area of interest on the machinable surface of a part, which can provide a unique set of removable volumes from part. Feature description in standard for exchange of product (STEP) AP224 is an efficient neutral format for the development of feature based process planning. Process planning information of features can be converted to numerical control (NC) code to have complete manufacturing information of part. STEP-NC code provides an efficient manufacturing information model compared to G-M codes. In this work, an interface is developed for extraction of feature information available in AP224 (AIM) format and the ruled-based approach is used to select different process planning parameters. A graphical user interface (GUI) is developed for the interface for displaying features information as represented in AP224 file. Furthermore, the interface generates STEP-NC code in AP238 format. The developed interface has three modules. 1) Module I: Reading interface for STEP AP224 file and development of GUI. 2) Module II: Selection of feature based process planning parameters. 3) Module III: Writing interface for STEP-NC (AP238). The developed interface has been implemented in Java through Java standard data access interface (JSDAITM). The generated STEP-NC AP238 code for the test part has been successfully simulated on STEP-NC MachineTM, an AP238 simulator. This article also provides an in-depth view of application interpreted model (AIM) representation format of STEP for AP224 and AP238.
Face recognition (FR) is a practical application of pattern recognition (PR) and remains a compelling topic in the study of computer vision. However, in real-world FR systems, interferences in images, including illumination condition, occlusion, facial expression and pose variation, make the recognition task challenging. This study explored the impact of those interferences on FR performance and attempted to alleviate it by taking face symmetry into account. A novel and robust FR method was proposed by combining multi-mirror symmetry with local binary pattern (LBP), namely multi-mirror local binary pattern (MMLBP). To enhance FR performance with various interferences, the MMLBP can 1) adaptively compensate lighting under heterogeneous lighting conditions, and 2) generate extracted image features that are much closer to those under well-controlled conditions (i.e., frontal facial images without expression). Therefore, in contrast with the later variations of LBP, the symmetrical singular value decomposition representation (SSVDR) algorithm utilizing the facial symmetry and a state-of-art non-LBP method, the MMLBP method is shown to successfully handle various image interferences that are common in FR applications without preprocessing operation and a large number of training images. The proposed method was validated with four public data sets. According to our analysis, the MMLBP method was demonstrated to achieve robust performance regardless of image interferences.
In this paper, the problem of hybrid model predictive control (HMPC) strategy based on fuzzy supervisor for piecewise autoregressive with exogenous input (PWARX) models is addressed. We first represent the nonlinear behavior of the system with a PWARX model. Then, we transform the obtained PWARX model into a mixed logical dynamic (MLD) model in order to apply the proposed predictive control which is able to stabilize such systems along desired reference trajectories while satisfying operating constraints. Finally, we propose to introduce a fuzzy supervisor allowing the readjustment of the HMPC tuning parameters in order to maintain the desired performance. Simulation and experimental results are presented to illustrate the effectiveness of the proposed approach.
This work proposes a soft sensor based on a phenomenological model for online estimation of the density and viscosity of a slurry flowing through a pipe-and-fittings assembly (PFA). The model is developed considering the conservation principle applied to mass and momentum transfer and considering frictional energy losses to include the variables directly affecting slurry properties. A reported proposal for state observers with unknown inputs is used to develop the first block of the observer structure. The second block is constructed with two options for evaluating slurry viscosity, generating two possible estimator structures, which are tested using real data. A comparison between them indicates different uses and capabilities according to available process information.
A Survey on 3D Visual Tracking of Multicopters
Qiang Fu, Xiang-Yang Chen, Wei He
Toolpath Interpolation and Smoothing for Computer Numerical Control Machining of Freeform Surfaces: A Review
Deep Learning Based Hand Gesture Recognition and UAV Flight Controls
Bin Hu, Jiacun Wang