Volume 14, Number 2, 2017

Display Method:
Review
A Survey on Deep Learning-based Fine-grained Object Classification and Semantic Segmentation
Bo Zhao, Jiashi Feng, Xiao Wu, Shuicheng Yan
2017, vol. 14, no. 2, pp. 119-135, doi: 10.1007/s11633-017-1053-3
Abstract:
The deep learning technology has shown impressive performance in various vision tasks such as image classification, object detection and semantic segmentation. In particular, recent advances of deep learning techniques bring encouraging performance to fine-grained image classification which aims to distinguish subordinate-level categories, such as bird species or dog breeds. This task is extremely challenging due to high intra-class and low inter-class variance. In this paper, we review four types of deep learning based fine-grained image classification approaches, including the general convolutional neural networks (CNNs), part detection based, ensemble of networks based and visual attention based fine-grained image classification approaches. Besides, the deep learning based semantic segmentation approaches are also covered in this paper. The region proposal based and fully convolutional networks based approaches for semantic segmentation are introduced respectively.
Research Article
PLS-CCA Heterogeneous Features Fusion-based Low-resolution Human Detection Method for Outdoor Video Surveillance
Hong-Kai Chen, Xiao-Guang Zhao, Shi-Ying Sun, Min Tan
2017, vol. 14, no. 2, pp. 136-146, doi: 10.1007/s11633-016-1029-8
Abstract:
In this paper, we focus on low-resolution human detection and propose a partial least squares-canonical correlation analysis (PLS-CCA) for outdoor video surveillance. The analysis relies on heterogeneous features fusion-based human detection method. The proposed method can not only explore the relation between two individual heterogeneous features as much as possible, but also can robustly describe the visual appearance of humans with complementary information. Compared with some other methods, the experimental results show that the proposed method is effective and has a high accuracy, precision, recall rate and area under curve (AUC) value at the same time, and offers a discriminative and stable recognition performance.
Method for Visual Localization of Oil and Gas Wellhead Based on Distance Function of Projected Features
Ying Xie, Xiang-Dong Yang, Zhi Liu, Shu-Nan Ren, Ken Chen
2017, vol. 14, no. 2, pp. 147-158, doi: 10.1007/s11633-017-1063-1
Abstract:
A localization method based on distance function of projected features is presented to solve the accuracy reduction or failure problem due to occlusion and blurring caused by smog, when dealing with vision based localization for target oil and gas wellhead (OGWH). Firstly, the target OGWH is modeled as a cylinder with marker, and a vector with redundant parameter is used to describe its pose. Secondly, the explicit mapping relationship between the pose vector with redundant parameter and projected features is derived. Then, a 2D-point-to-feature distance function is proposed, as well as its derivative. Finally, based on this distance function and its derivative, an algorithm is proposed to estimate the pose of target OGWH directly according to the 2D image information, and the validity of the method is verified by both synthetic data and real image experiments. The results show that this method is able to accomplish the localization in the case of occlusion and blurring, and its anti-noise ability is good especially with noise ratio of less than 70%.
A New Monocular Vision Measurement Method to Estimate 3D Positions of Objects on Floor
Ling-Yi Xu, Zhi-Qiang Cao, Peng Zhao, Chao Zhou
2017, vol. 14, no. 2, pp. 159-168, doi: 10.1007/s11633-016-1047-6
Abstract:
A new visual measurement method is proposed to estimate three-dimensional (3D) position of the object on the floor based on a single camera. The camera fixed on a robot is in an inclined position with respect to the floor. A measurement model with the camera's extrinsic parameters' such as the height and pitch angle is described. Single image of a chessboard pattern placed on the floor is enough to calibrate the camera's extrinsic parameters' after the camera's intrinsic parameters' are calibrated. Then the position of object on the floor can be computed with the measurement model. Furthermore, the height of object can be calculated with the paired-points in the vertical line sharing the same position on the floor. Compared to the conventional method used to estimate the positions on the plane, this method can obtain the 3D positions. The indoor experiment testifies the accuracy and validity of the proposed method.
Singularity Robust Path Planning for Real Time Base Attitude Adjustment of Free-floating Space Robot
Cheng Zhou, Ming-He Jin, Ye-Chao Liu, Ze Zhang, Yu Liu, Hong Liu
2017, vol. 14, no. 2, pp. 169-178, doi: 10.1007/s11633-017-1055-1
Abstract:
This paper presents a singularity robust path planning for space manipulator to achieve base (satellite) attitude adjustment and end-effector task. The base attitude adjustment by the movement of manipulator will save propellant compared with conventional attitude control system. A task-priority reaction null-space control method is applied to achieve the primary task of adjusting attitude and secondary task of accomplishing end-effector task. Furthermore, the algorithm singularity is eliminated in the proposed algorithm compared with conventional reaction null-space algorithm. And the singular value filtering decomposition is introduced to dispose the dynamic singularity, the unit quaternion is also introduced to overcome representation singularity. Hence, a singularity robust path planning algorithm of space robot for base attitude adjustment is derived. A real time simulation system of the space robot under Linux/RTAI (realtime application interface) is developed to verify and test the feasibility and reliability of the method. The experimental results demonstrate the feasibility of online base attitude adjustment of space robot by the proposed algorithm.
Quasi-inverse Pendulum Model of 12 DoF Bipedal Walking
M. Akhtaruzzaman, Amir A. Shafie, Raisuddin Md. Khan
2017, vol. 14, no. 2, pp. 179-190, doi: 10.1007/s11633-016-1023-1
Abstract:
This paper presents modeling of a 12-degree of freedom (DoF) bipedal robot, focusing on the lower limbs of the system, and trajectory design for walking on straight path. Gait trajectories are designed by modeling of center of mass (CoM) trajectory and swing foot ankle trajectory based on stance foot ankle. The dynamic equations of motion of the bipedal robot are derived by considering the system as a quasi inverted pendulum (QIP) model. The direction and acceleration of CoM movement of the QIP model is determined by the position of CoM relative to the centre of pressure (CoP). To determine heel-contact and toe-off, two custom designed switches are attached with heel and toe positions of each foot. Four force sensitive resistor (FSR) sensors are also placed at the plantar surface to measure pressure that is induced on each foot while walking which leads to the calculation of CoP trajectory. The paper also describes forward kinematic (FK) and inverse kinematic (IK) investigations of the biped model where Denavit-Hartenberg (D-H) representation and Geometric-Trigonometric (G-T) formulation approach are applied. Experiments are carried out to ensure the reliability of the proposed model where the links of the bipedal system follow the best possible trajectories while walking on straight path.
An Adaptive Full Order Sliding Mode Controller for Mismatched Uncertain Systems
Mondal Sanjoy, Ghommam Jawhar, Saad Maarouf
2017, vol. 14, no. 2, pp. 191-201, doi: 10.1007/s11633-017-1057-z
Abstract:
In this paper, an adaptive full order sliding mode (FOSM) controller is proposed for strict feedback nonlinear systems with mismatched uncertainties. The design objective of the controller is to track a specified trajectory in presence of significant mismatched uncertainties. In the first step the dynamic model for the first state is considered by the desired tracking signal. After the first step the desired dynamic model for each state is defined by the previous one. An adaptive tuning law is developed for the FOSM controller to deal with the bounded system uncertainty. The major advantages offered by this adaptive FOSM controller are that advanced knowledge about the upper bound of the system uncertainties is not a necessary requirement and the proposed method is an effective solution for the chattering elimination from the control signal. The controller is designed considering the full-order sliding surface. System robustness and the stability of the controller are proved by using the Lyapunov technique. A systematic adaptive step by step design method using the full order sliding surface for mismatched nonlinear systems is presented. Simulation results validate the effectiveness of the proposed control law.
A Sliding Mode Observer for Uncertain Nonlinear Systems Based on Multiple Models Approach
Hfaïedh Kaïs, Dahech Karim, Damak Tarak
2017, vol. 14, no. 2, pp. 202-212, doi: 10.1007/s11633-016-0970-x
Abstract:
This paper presents a method of state estimation for uncertain nonlinear systems described by multiple models approach. The uncertainties, supposed as norm bounded type, are caused by some parameters' variations of the nonlinear system. Linear matrix inequalities (LMIs) have been established in order to ensure the stability conditions of the multiple observer which lead to determine the estimation gains. A sliding mode gain has been added in order to compensate the uncertainties. Numerical simulations through a state space model of a real process have been realized to show the robustness of the synthesized observer.
Regional Stability of Positive Switched Linear Systems with Multi-equilibrium Points
Zhi Liu, Yu-Zhen Wang
2017, vol. 14, no. 2, pp. 213-220, doi: 10.1007/s11633-016-1003-5
Abstract:
This paper studies the regional stability for positive switched linear systems with multi-equilibrium points (PSLS-MEP). First, a sufficient condition is presented for the regional stability of PSLS-MEP via a common linear Lyapunov function. Second, by establishing multiple Lyapunov functions, a dwell time based condition is proposed for the regional stability analysis. Third, a suprasphere which contains all equilibrium points is constructed as a stability region of the considered PSLS-MEP, which is less conservative than existing results. Finally, the study of an illustrative example shows that the obtained results are effective in the regional stability analysis of PSLS-MEP.
An Adaptive Regulation Problem and Its Application
Yuan Jiang, Ji-Yang Dai
2017, vol. 14, no. 2, pp. 221-228, doi: 10.1007/s11633-015-0900-3
Abstract:
This paper studies an adaptive regulation problem for the modified FitzHugh-Nagumo neuron model under external electrical stimulation. We first present the solution of the global robust output regulation problem for output feedback system with an uncertain exosystem which models the external electrical stimulation with unknown frequency and amplitude. Then, we show that the robust control problem for the modified FitzHugh-Nagumo neuron model can be formulated as the global robust output regulation problem and the solvability conditions for the output regulation problem for the modified FitzHugh-Nagumo neuron model are all satisfied. Then, we apply the obtained output regulation result to constructing an output feedback control law for the modified FitzHugh-Nagumo neuron model to achieve global stability of the closed-loop system in the presence of uncertain parameters' and external stimulus. An example is given to show that the proposed algorithm can completely reject the external electrical stimulation.
Dynamics and Stability of Potential Hyper-networked Evolutionary Games
Ting Liu, Yuan-Hua Wang, Dai-Zhan Cheng
2017, vol. 14, no. 2, pp. 229-238, doi: 10.1007/s11633-017-1056-0
Abstract:
This paper considers the modeling and convergence of hyper-networked evolutionary games (HNEGs). In an HNEG the network graph is a hypergraph, which allows the fundamental network game to be a multi-player one. Using semi-tensor product of matrices and the fundamental evolutionary equation, the dynamics of an HNEG is obtained and we extend the results about the networked evolutionary games to show whether an HNEG is potential and how to calculate the potential. Then we propose a new strategy updating rule, called the cascading myopic best response adjustment rule (MBRAR), and prove that under the cascading MBRAR the strategies of an HNEG will converge to a pure Nash equilibrium. An example is presented and discussed in detail to demonstrate the theoretical and numerical results.