Xinhe Xu

EEG Analysis of Alcoholics and Controls Based on Feature Extraction

In this paper, we investigated EEG feature in the alcoholics and the controls. Principle component analysis was applied to preprocess the original data to reduce the dimensions of EEG. Wavelet transform decomposed the EEGs into five corresponding frequency bands. Power spectrum was estimated in each band. By comparing the power spectrum of the alcoholics and controls in five major frequency bands, we found that increased theta and delta power can be observed obviously in the alcoholics, while alpha power of alcoholics was significantly decreased. In the gamma and beta bands, the power of controls was a little greater than that of alcoholics. The results of the experimental analysis are valuable to alcohol detection and have significance to the brain computer interface research

Optimization of Controller Parameters Based on the Improved Genetic Algorithms

Genetic algorithm with many performance indexes is proposed. It can seek an optimal application in the controller parameters. And the concrete implementation method of the improved genetic algorithm is discussed. In the improved genetic algorithm, the constructed formula of many performance indexes is also given and the aspects of formulation have been discussed. Genetic algorithm with many performance indexes has many representative ways, one of which is involved in simulation. The selection of the performance index has considered the coefficient of weight in order to display the importance of every performance index. Simulation results have proved new method have better parameter seek optimally ability than the genetic algorithm with single performance index

A detection method of infrared image small target based on order morphology transformation and image entropy difference

In this paper a novel approach of infrared image small target detection in background of sky and sea is put forward. Firstly, method of line-average-value- subtraction and order morphology filter are used to restrain correlative background and obtain candidate targets; then two different target detection methods are proposed and used to detect targets. One of them is that order morphological edge detection operators constructed are adopted to wipe off noise and false target while edge of target was extracted; another is that using image entropy difference to detect the discontinuity between targets and background. The experimental results show that the techniques presented in this paper are valid.

Face detection based on skin color

Face detection plays an important role in many applications such as surveillance, human computer interface, face recognition, and face image database management. This paper proposes a face detection algorithm for color images, which is based on an adaptive threshold and chroma chart that shows probability of skin colors. The experiments with variety of images show that the method proposed here is efficient and useful.

Design and Implementation of Coordinated Control Strategy for Biped Robot with Heterogeneous Legs

To meet the complicated control requirement of biped robot with heterogeneous legs (BRHL), the master-slave dual-leg coordinated control strategy was adopted to simplify its planning scheme, of which the key of coordination is gait tracking control. Bionic knee joint enables the robot to walk more like human, but its modeling is complex and the control of bionic leg to track artificial leg is difficult. The P-type open/closed-loop iterative learning control (ILC) is therefore put forward to combine together the merits of open-loop and closed-loop ILC without a model to rely on so as to fit the control of gait tracking for complicated biped robot system. Experimental results demonstrate the correctness and effectiveness of the proposed coordinated control scheme.

Modeling and control with hysteresis and creep of ionic polymer-metal composite (IPMC) actuators

Hysteresis and creep hinder the effective use of IPMC in sensors and actuators. This paper proposes a hybrid model that can precisely portray hysteresis and creep in piezoelectric actuators, which is constructed by a Preisach operator with a piecewise uniform density function and creep operator. Then, the corresponding inverse models for both hysteresis and creep are developed. It studies online recursive identification of hysteresis and creep drift. Based on the obtained models, a method for simultaneous compensation of the hysteresis and creep of piezoelectric actuator is applied to the control of system nonlinearities. Simulation and experimental results based on a IPMC actuator are provided to illustrate the proposed approach. The result verified the validity of the model and effectiveness of the controller.

Research on gait planning of artificial leg based on central pattern generator

Biped robot with heterogeneous legs (BRHL) is a novel robot model, which consists of an artificial leg and an intelligent bionic leg. The artificial leg is used to simulate the amputeepsilas healthy leg and the bionic leg works as the intelligent artificial limb. To describe the present gait of the healthy leg and make intelligent bionic leg follow the walking of artificial leg in all phases is the target of BRHLpsilas research. So gait planning of artificial leg is the emphasis of BRHLpsilas research. This paper uses the model of central pattern generator (CPG) in the research of artificial legpsilas gait planning from the point of biology. To obtain natural and robust walking pattern, genetic algorithm is used to optimize parameters of the CPG network model and the fitness function is formulated based on zero moment point (ZMP). Simulation results testify the feasibility of this method.

Study and application of medical image visualization technology

In medical imaging, many applications require visualization and analysis of three-dimensional (3D) objects. Visualization is the process of exploring, transforming, and view data as images to gain understanding and insight into the data, which requires fast interactive speed and high image quality. In this paper, we describe the key techniques in medical image visualization. In order to improve ray casting rendering speed, a synthetically accelerated algorithm is proposed. Firstly, rendering algorithms are fully studied and compared. Secondly, proximity clouds algorithm has been selected and extended to continuous ray casting. Finally, the accelerated algorithm based on ray coherence has been realized. The experimental results on 3D medical image reconstruction are given, which show the medical image visualization technology has provided a powerful technology base for computer-aided diagnosis, virtual surgery and e-learning in medicine field.

Terrain Identification of Intelligent Bionic Leg Based on Ground Reaction Force

Biomedical research indicates that human gait is considerable different when walking on different terrain. In this paper, terrain identification scheme for an intelligent bionic leg (IBL) is presented. Terrain identification system is modeled by the procedure neural networks (PNNs) and is used to identify the current terrain type based on the ground reaction force (GRF) information. According to identification result, damping moment of magneto-rheological (MR) damper mounted on knee joint of IBL is adjusted to accommodate changing conditions. This paper presents the PNNs model and calculates the related arithmetics. Terrain identification results prove the rationality and feasibility of the proposed scheme.

Algorithm of image enhancement based on order morphology filtering and image entropy difference

According to correlative conception and properties of order morphology transformation, a non-linear weighted mean value filter is constructed to remove high frequence noise such Gaussian and impulse noise. Then an approximated midpoint value filter, which can reject uniform distribution noise is given by improving the weighted mean value filter. Based on the above, a new image enhancement algorithm is proposed. In this algorithm, weighted mean value of the image about structuring elements of different directions is calculated and used to identify the edge of the image; local average value and image entropy difference is applied to control enhancement coefficients. So the target and edge of the image are elevated while high frequency noise of the image is restrained. A comparison of average value, standard deviation and image entropy between the enhanced image and its original illustrates that contrast of the image is also improved.