摘要：
Spatial position consistency and occlusion consistency are two important problems in augmented reality systems. In this paper, we proposed a novel method that can address the registration problem and occlusion problem simultaneously by using an RGB-D camera. First, to solve the image alignment errors caused by the imaging mode of the RGB-D camera, we developed a depth map inpainting method that combines the FMM and RGB-D information. Second, we established an automatic method to judge the close-range mode based on the depth histogram to solve the registration failure problem caused by hardware limitations. In the close-range mode, the registration method combining the fast ICP and ORB was adopted to calculate the camera pose. Third, we developed an occlusion handling method based on the geometric analysis of the scene. Several experiments were performed to validate the performance of the proposed method. The experimental results indicate that our method can obtain stable and accurate registration and occlusion handling results in both the close-range and non-close-range modes. Moreover, the mutual occlusion problem can be handled effectively, and the proposed method can satisfy the real-time requirements of augmented reality systems.
作者机构：
[Luo Da-Xiong; Shu, Chen; Song, Xu; Min, Ye Jun] Cent China Normal Univ, Sch Comp, Wuhan 430070, Hubei, Peoples R China.;[Feng, Wang Zhi] Cent China Normal Univ, Sch Educ Informat Technol, Wuhan 430070, Hubei, Peoples R China.
通讯机构：
[Song, Xu] C;Cent China Normal Univ, Sch Comp, Wuhan 430070, Hubei, Peoples R China.
会议名称：
9th International Conference of Information and Communication Technology [ICICT]
会议时间：
JAN 11-13, 2019
会议地点：
Nanning, PEOPLES R CHINA
会议主办单位：
[Song, Xu;Min, Ye Jun;Luo Da-Xiong;Shu, Chen] Cent China Normal Univ, Sch Comp, Wuhan 430070, Hubei, Peoples R China.^[Feng, Wang Zhi] Cent China Normal Univ, Sch Educ Informat Technol, Wuhan 430070, Hubei, Peoples R China.
摘要：
The short text in the online learning community is an important source of data in learning analysis. Therefore, the quality of the short text has a significant impact on the study of learning analysis. Due to the large amount of text data in the learning community, manual detection and repair will cost too much. This paper proposes a text detection and repair framework based on an online learning community. It aims to automatically detect and repair various types of semantic errors and grammatical errors that exist in online learning community short texts. The framework utilizes existing text error detection and repair algorithms and integrates them effectively to form a comprehensive detection and repair algorithm. In this paper, the validity of the framework is verified through experiments on the constructed data set. The experimental results show that the framework has high accuracy in automatically detecting and repairing text errors. (C) 2019 The Authors. Published by Elsevier Ltd.
摘要：
对话流所隐含的信息包括了学习者对所学课程内容的掌握程度和关注点,分析这些对话流对预测学习者的成绩,以支持教师提前对潜在成绩不良的学生进行及时干预有着重要意义.提出了一种基于对话流的学习者成绩等级预测算法ARPDF(Achievement Rank Prediction based on Dialogue Flow),首先采集对话流,通过对话流划分、对话状态矩阵生成实现了对该对话流的分析以获取到学习小组的对话状态矩阵;在此基础上,通过基于LSTM的预测模型获得学习小组学习者的成绩等级.在本文所提方法的基础上进行了实验,其结果表明了该算法是有效的.
期刊：
Lecture Notes on Data Engineering and Communications Technologies,2018年 6: 160-166 ISSN：2367-4512
通讯作者：
Wu, Minghu
作者机构：
[Zhao, Nan; Rao, Zheheng; Wu, Minghu; Liu, Min; Zeng, Chunyan] Hubei Univ Technol, Hubei Key Lab High Efficiency Utilizat Solar Ener, Wuhan, Hubei, Peoples R China.;[Zhao, Nan; Rao, Zheheng; Wu, Minghu; Liu, Min; Zeng, Chunyan] Hubei Univ Technol, Hubei Collaborat Innovat Ctr High Efficiency Util, Wuhan 430068, Hubei, Peoples R China.;[Wang, Zhifeng] Cent China Normal Univ, Sch Educ Informat Technol, Wuhan 430079, Hubei, Peoples R China.
通讯机构：
[Wu, Minghu] H;Hubei Univ Technol, Hubei Key Lab High Efficiency Utilizat Solar Ener, Wuhan, Hubei, Peoples R China.;Hubei Univ Technol, Hubei Collaborat Innovat Ctr High Efficiency Util, Wuhan 430068, Hubei, Peoples R China.
会议名称：
5th International Conference on Emerging Internetworking, Data and Web Technologies (EIDWT)
会议时间：
JUN 10-11, 2017
会议地点：
Wuhan, PEOPLES R CHINA
会议主办单位：
[Rao, Zheheng;Zeng, Chunyan;Zhao, Nan;Liu, Min;Wu, Minghu] Hubei Univ Technol, Hubei Key Lab High Efficiency Utilizat Solar Ener, Wuhan, Hubei, Peoples R China.^[Rao, Zheheng;Zeng, Chunyan;Zhao, Nan;Liu, Min;Wu, Minghu] Hubei Univ Technol, Hubei Collaborat Innovat Ctr High Efficiency Util, Wuhan 430068, Hubei, Peoples R China.^[Wang, Zhifeng] Cent China Normal Univ, Sch Educ Informat Technol, Wuhan 430079, Hubei, Peoples R China.
会议论文集名称：
Lecture Notes on Data Engineering and Communications Technologies
摘要：
This paper presents a new deep network (non - very deep network) with composite residual for handwritten character recognition. The main network design is as follows: (1) Introduces an unsupervised FCM clustering algorithm to preprocess the experimental data. (2) By exploiting a composite residual structure the multilevel shortcut connection is proposed which is more suitable for the learning of residual. (3) In order to solve the problem of over-fitting and time-consuming for training the network parameters, a dropout layer is added after the completion of all convolution operations of each extended nonlinear residual kernel. Comparing with general deep network structures of same deep on handwritten character MNIST database, the proposed algorithm shows better recognition accuracy and higher recognition efficiency.
作者机构：
[Ye Jun-min; Xu Song; Luo Da-Xiong; Huang Peng-Wei; Xu Chen] School of Computer, Central China Normal University, Wu Han, 430000, China;[Wang Zhi-Feng] School of Educational Information Technology, Central China Normal University, Wu Han, 430000, China
通讯机构：
School of Computer, Central China Normal University, Wu Han, China
摘要：
Aiming at the problem of "learning defiance" and "information overload" brought by educating big data to learners, this paper proposes an online learning community personalized learning path recommendation algorithm based on ant colony algorithm: in terms of computing pheromone, it combines individuality. Based on the characteristics of the learning path, a learning path scoring method based on multi-factor fuzzy evaluation is proposed to quantify the learning path evaluation as a score to solve the problem that it is difficult for the subjective score to accurately represent the pheromone concentration; in terms of pheromone updating rules, The introduction of pheromone restriction intervals avoids the problems associated with excessive or small learning path pheromone concentration in global updating; in the calculation of the selection probability of local search, the positive and negative feedback effects of pheromones can be better used. Search for a local optimal solution. The related experiments show that this algorithm can effectively solve the recommendation of the personalized learning path of the online learning community.
摘要：
Although the accuracy of handwritten character recognition based on deep networks has been shown to be superior to that of the traditional method, the use of an overly deep network significantly increases time consumption during parameter training. For this reason, this paper took the training time and recognition accuracy into consideration and proposed a novel handwritten character recognition algorithm with newly designed network structure, which is based on an extended nonlinear kernel residual network. This network is a non-extremely deep network, and its main design is as follows:(1) Design of an unsupervised apriori algorithm for intra-class clustering, making the subsequent network training more pertinent; (2) presentation of an intermediate convolution model with a pre-processed width level of 2;(3) presentation of a composite residual structure that designs a multi-level quick link; and (4) addition of a Dropout layer after the parameter optimization. The algorithm shows superior results on MNIST and SVHN dataset, which are two character benchmark recognition datasets, and achieves better recognition accuracy and higher recognition efficiency than other deep structures with the same number of layers.
摘要：
In the online learning environment, identifying learners' behaviors in the learning process can help them improve their learning effect autonomously. Firstly, we use K-Means algorithm to cluster the learner's help-seeking behavior data to get the classification label of the learner's help-seeking behavior. Secondly, we use the t-distributed Stochastic Neighbor Embedding(T-sne) algorithm to reduce the dimension of the data to visualize the clustering result. Finally, the learner's help-seeking behavior data and the help-seeking behavior classification labels are used as training data to train the Naive Bayesian model so as to automatically obtain the help-seeking behavior classification for the data generated by the new learner. Via the analysis and processing of the help-seeking behavior data using the method proposed in this paper, it shows that this method can effectively find online learners' help-seeking behavior classifications.
摘要：
With the rapid development of online education, virtual learning community has a great impact on people's learning. Here we focus on the dynamics evolution law of collaborative behavior, characterizing the features of collaborative behaviors among learners in virtual learning community. We construct a learners' collaboration network model in virtual learning community. We analyze important statistics of the network such as the degree distribution, the average path length, the clustering coefficients. On this basis, we describe the network features and study the topology characteristics of the collaborative relationship network deeply. The experiments show that the learners' collaboration network model could depict the characteristics of virtual learning community and the characteristics of the collaborative behavior among learners more accurately. The research could enrich the existing network model construction theory, and provide the network structure foundation for the study of collaborative relationship between learners in virtual learning community.
摘要：
Obtaining the correct occlusion relationship between real objects and virtual objects is vital for improving augmented reality technology. In this paper, we propose a novel occlusion handling method using moving volume and ray casting techniques. Our method is divided into two steps. In the first step, we obtain the volume of the corresponding physical space and arbitrarily move the volume to extend the reconstruction area. In the second step, we calculate the 3D coordinates of each pixel in the scene and re-project the rendered objects to the same 3D coordinates system. Correct occlusion relationships are obtained by comparing the z coordinates of real and virtual objects. Several experiments are performed to validate the performance of the proposed method. The experimental results indicate that our method can correctly and rapidly handle occlusion.
摘要：
In today's Internet+ era, more and more e-learning platforms have appeared in the eyes of the public. The emergence of big data also points a development path for the learning platform. By analyzing the latest research progress and existing problems of many learning platforms, an online learning system with learning feedback is proposed. The system consists of a learning subsystem and a feedback subsystem. The learning subsystem is designed based on the basic needs of students. The feedback subsystem uses clustering algorithms and some calculation formulas to analyze student learning behaviors and display them in the summary report to the teacher. Teachers can adjust the teaching plan according to the summary report.
