In order to meet the application demand for informatization, intellectualization, real-time and visualization of chemical plan, this paper proposes a real-time monitoring project based on the Internet of Things and visualization technology. In the project, with the backing of data acquisition, transmission, storage, analysis and display in cloud architecture and full use of ZigBee, a new wireless sensor network, a real-time visual monitoring system, which can be run on multiple platforms, is constructed through synthetic analysis of the actual monitoring requirements and the application characteristics. This paper elaborates the architecture, the function-constructing project and 3D visual platform of the monitoring system and makes an exploratory study of the real-time monitoring in a new era.

Classifying electroencephalogram (EEG) signals is one of the most important issues on motor imagery-based Brain computer interfaces (BCIs). Typically, such classification has been performed using a small training dataset To date, most of the classification of the algorithms were proposed for large samples. In this paper, a combination of Rayleigh coefficient maximization and graph-based method was developed to classify EEG signals with small training dataset. The Rayleigh coefficient maximization was adopted to obtain the projection directions, which extract discriminating features from the preprocessed dataset. Next, both training and testing features are applied to construct an affinity matrix, and then both affinity matrix and all label information are applied to train a classifier based on graph-based semi-supervised method. In this approach, both labeled and unlabeled samples are used for training a classifier. Hence it can be used in small training data case. Finally, a new iteration mechanism is applied to update the training data set. And the experiment results on BCI competition III dataset IVa show that the classification accuracy using our method was higher than using CSP (common spatial pattern) and support vector machine (SVM) method in all subjects with different size of training dataset We used an eightfold cross-validation on this dataset, and the results show a good stability of our algorithm.

This paper adopts ordinary CCD cameras and IR filters to constitute IR cameras for a marker-based quadri-ocular tracking system. However, the IR filters will affect the accuracy of the tracking system. Therefore, this paper analyses the refraction of IR filters and finds IR filters will bring about the biggish radial distortion on imaging process. And then, this paper directly uses Zhang’s technique to calibrate the parameters of IR cameras in infrared environment by adjusting the imaging conditions of calibration. To improve the accuracy and satisfy the real-time requirement of the tracking system, a real-time method to correct the radial distortion is proposed in this paper. The experiments show that the speed of the correction method is satisfactory, and the accuracy of tracking system is effectively improved by the method.

Camera is widely used for 3D-Reconstruction and Recognition where a fundamental task is point correspondence that finds the corresponding points captured by different cameras from the same 3D-point. The approaches based on epipolar constraint are very effective for the task. However, they set the threshold of epipolar constraint by empirical method, and their complexity and computational cost will quickly increase with increasing the number of cameras. This paper proposes a novel two-step strategy using epipolar constraint regardless of the number of cameras in a uniform way. The strategy uses a statistic method to set the threshold, and decomposes the task of point correspondence in a multi-ocular system into the task of point correspondence in several binocular systems by transitivity constraint. The experiment and theoretical analyses indicate our approach is better than existing methods on complexity and computational cost.

Recent years, many vision-based vehicle detection methods have been proposed. But just a few of them paid attention to vehicle type classification in a real-time video, which is an important part in intelligent traffic system. The Haar features and Histograms of Oriented Gradients (HOG) features have been proposed as useful features for vehicle detection respectively. In this paper, we propose a method based on combined the Haar features and HOG features to detect vehicles in videos and classify them into two types. The experiment results show that this method can classify and detect the vehicles in multi-orientations with good classification results, and it verified the effectiveness of the propose method with real vehicle samples.

The Pedestrian detection using Histograms of Oriented Gradients (HOG) is the most popular method to detect a human from a picture. However, it, calculates the HOG description, will cost too much time and can't meet the real-time request for detecting pedestrian from the video surveillance system. In this paper we present a novel algorithm for detecting a human from a video. Firstly, The improved approach of Vibe follows a new background model using the temporal information, and present a new post-processing method for expanding the outlines of the foreground objects and then extract the foreground objects zone. Secondly, calculating the HOG feature of the extracted zone, and then send into the SVM classifier which has been trained to judge where is pedestrian or not. The combination algorithm, the improved Vibe and the HOG pedestrian detection, can save the processing time and the simulation results show that the proposed algorithm, compared with the traditional pedestrian detection algorithm, can detect pedestrian more accuracy and efficiency and its optimization ability is stronger.

Traditional optical flow method can not meet the requirement of real-time property due to huge computation workload when extracting the interested region.In light of this,we propose an algorithm for extracting interested region which is based on corner moving constraints.First,the new algorithm uses Harris algorithm to detect corners on the video image,by pre-processing the corner area to extract the corner area of the foreground.Secondly,on this basis the optical flow method is made use of to establish the optical flow field of corner area,by setting up motion constraint and threshold treatment to extract the foreground of moving target.Experimental results of the algorithm show that the algorithm can accurately extract the interested region,it is strong in anti-jamming and can meet real-time requirement.

This paper proposes an improved method of Vibe (Visual background extractor) based on Opencv (Open Source Computer Vision Library) on Android operating system. It briefly introduced the processing procedures which build the development environment with Opencv4 Android and implement this application program by using the Android framework. The step of using JNI (Java Native Interface) to call the native functions of OpenCV is explained and the using of Android SDK (Software Development Kit) and NDK (Native Development Kit) to generate shared library are also mentioned. Besides, this paper achieves some improvements of Vibe, a motion detection method, involving shadow elimination and noise inhibition. Android is the most popular system on mobile platform, more and more mobile devices are equipped with Android. So it is worthy to implement the motion detection computer vision algorithm based on Android Opera System. And the experiment results show the effectiveness of the proposed method based on Android system.

In this study, a novel artificial biofeedback system based on the transcutaneous electrical nerve stimulation and pattern recognition of surface electromyography(sEMG) signals is designed for the rehabilitation treatment. This system is composed of hardware circuit of sEMG acquisition, surface Agcl electrodes, electrical nerve stimulator and relevant software. The main purpose of the system is to cure the nerve and muscle disease by biofeedback intelligent technology instead of physicians, that is, by means of feature extraction and classification of sEMG, the system can identify three different state (sensory, motorial, painful) and the fatigue state of the muscle, then according to above discrimination results to control the output of the stimulator automatically. In this paper, Firstly, a surface electromyographic signal acquisition circuit and signal processing interface based MFC are developed and designed. Secondly, the AR(Auto-Regressive)and WT(wavelet transform) are adopted for signal feature extraction, then extracted feature vectors are feed to the SVM(support vector machine) classifier. Finally, according to the discrimination results to regulate the output of the stimulator. Experiments verify the effectiveness of the system.

Moving shadow detection and removal are key steps for motion detection algorithm. But by far most of the traditional methods, relying solely on single information, can not eliminate shadow effectively. This paper describes a mixed approach to deal with the shadow of the foreground objects from video surveillance. Firstly, a new description of local texture operator-LMTO (Local Match Texture Operator) is adopted; it can be an excellent capability in describing the local texture information. Secondly, we propose a new determination mechanism which combines the luminance, texture information and color-ratio of the pixels. Experimental results show that the proposed algorithm is efficient and robust.

Clustering is one of the building blocks of modern data analysis such as image processing, data mining, and pattern recognition. Path-based spectral clustering is an important approach for clustering, which has delivered impressive results in some challenging tasks. However this algorithm has huge time costing due to the number of paths will dramatically rise with the increase of dataset size. For this problem, this paper proposes a novel spectral clustering method that utilizes the density of dataset to limit the scope of paths instead of finding all the paths. Experiments on synthetic as well as real world data sets and the run time of algorithms demonstrate that the proposed method outperforms the path-based algorithm.

The ant colony algorithm is a mature and effective method to solve the problem of optimizing shortest path, which is one of the key technologies for robot navigation and path planning. But the algorithm often fails into precocity easily and can't get the global best result. This paper proposes an improved ant colony optimization algorithm by dynamically adjusting ant number. The main idea of this algorithm is that only the part of the ants passing the shorter path is allowed to release pheromone and update the total ant number randomly or fixedly in algorithm iterative process. So, the improved algorithm can increase the randomness in the search and improve global search ability. To verify the performance of this algorithm, this paper uses the improved algorithm to solve Chinese Traveling Salesmen Problem. The simulation results show that compared with the traditional ant colony algorithm, the improved ant colony algorithm is easier to find the optimal solution, and its optimization ability is stronger.

Functional Electrical Stimulation (FES) has already been used in many aspects of rehabilitation medicine as a treatment method. According to the functional electrical stimulation, we present an electrical muscle simulator which can adjust the amplitude, frequency and wave of electrical impulses by button and make the treatment more effective. In this paper, we will introduce the basic principle of the functional electrical stimulation and the basic parameters of stimulation waveform, as well as we will present a framework of this device. There are two crucial components should be elaborately designed, the hardware system and the software system. Moreover, the analysis of the electrical signal is mentioned in this paper last.

In video surveillance, there are a variety of random disturbances on moving target detection, such as trees sway, camera shake. In order to eliminate these disturbances, this paper presents a novel moving object detection algorithm based on the properties of corner points. Firstly, this paper uses Harris algorithm to detect corner on the video image, and then proposes a novel indicator, Inter-frame Regional Corners Difference, to select the candidate grids of foreground. In this way, the static and slight shaking background is recognized. Secondly, this paper makes use of Horn and Schunck's optical flow algorithm to build the optical flow field of grids that are the candidates of foreground, and extracts the moving target by some velocity constraint of the horizontal and vertical direction. In virtue of the different dynamic properties of background and moving target, the dynamic background can be eliminated. The experimental results show that our algorithm can accurately extract moving target and can meet the needs of real-time processing with strong anti-jamming.

研究经不同条件热处理的Cu-8.9Al-12.4Mn-0.07B低温形状记忆合金的结构及相变行为,发现B的添加能细化晶粒,并抑制室温时效过程中Cu9Al4（γ2）相的析出。