Online learning and fusion of orientation appearance models for robust rigid object tracking

We introduce a robust framework for learning and fusing of orientation appearance models based on both texture and depth information for rigid object tracking. Our framework fuses data obtained from a standard visual camera and dense depth maps obtained by low-cost consumer depth cameras such as the Kinect. To combine these two completely different modalities, we propose to use features that do not depend on the data representation: angles. More specifically, our framework combines image gradient orientations as extracted from intensity images with the directions of surface normals computed from dense depth fields. We propose to capture the correlations between the obtained orientation appearance models using a fusion approach motivated by the original Active Appearance Models (AAMs). To incorporate these features in a learning framework, we use a robust kernel based on the Euler representation of angles which does not require off-line training, and can be efficiently implemented online. The robustness of learning from orientation appearance models is presented both theoretically and experimentally in this work. This kernel enables us to cope with gross measurement errors, missing data as well as other typical problems such as illumination changes and occlusions. By combining the proposed models with a particle filter, the proposed framework was used for performing 2D plus 3D rigid object tracking, achieving robust performance in very difficult tracking scenarios including extreme pose variations.     

Automatic annotation of tennis games: An integration of audio,vision, and learning

Fully automatic annotation of tennis game using broadcast video is a task with a great potential but with enormous challenges. In this paper we describe our approach to this task, which integrates computer vision, machine listening, and machine learning. At the low level processing, we improve upon our previously proposed state-of-the-art tennis ball tracking algorithm and employ audio signal processing techniques to detect key events and construct features for classifying the events. At high level analysis, we model event classification as a sequence labelling problem, and investigate four machine learning techniques using simulated event sequences. Finally, we evaluate our proposed approach on three real world tennis games, and discuss the interplay between audio, vision and learning. To the best of our knowledge, our system is the only one that can annotate tennis game at such a detailed level.     

An improved global-best harmony search algorithm for faster optimization

In this paper, an improved global-best harmony search algorithm, named IGHS, is proposed. In the IGHS algorithm, initialization based on opposition-based learning for improving the solution quality of the initial harmony memory, a new improvisation scheme based on differential evolution for enhancing the local search ability, a modified random consideration based on artificial bee colony algorithm for reducing randomness of the global-best harmony search (GHS) algorithm, as well as two perturbation schemes for avoiding premature convergence, are integrated. In addition, two parameters of IGHS, harmony memory consideration rate and pitch adjusting rate, are dynamically updated based on a composite function composed of a linear time-varying function, a periodic function and a sign function in view of approximate periodicity of evolution in nature. Experimental results tested on twenty-eight benchmark functions indicate that IGHS is far better than basic harmony search (HS) algorithm and GHS. In further study, IGHS has also been compared with other eight well known metaheuristics. The results show that IGHS is better than or at least similar to those approaches on most of test functions.     

基于蚁群聚集信息素的半监督文本分类算法

半监督文本分类中已标记数据与未标记数据分布不一致,可能导致分类器性能较低。为此,提出一种利用蚁群聚集信息素浓度的半监督文本分类算法。将聚集信息素与传统的文本相似度计算相融合,利用Top-k策略选取出未标记蚂蚁可能归属的种群,依据判断规则判定未标记蚂蚁的置信度,采用随机选择策略,把置信度高的未标记蚂蚁加入到对其最有吸引力的训练种群中。在标准数据集上与朴素贝叶斯算法和EM算法进行对比实验,结果表明,该算法在精确率、召回率以及F1度量方面都取得了更好的效果。     

纹理豪斯多夫距离在人脸识别算法中的应用

很多经典的人脸识别方法难以适应姿势变化及人脸错位的情形。为了解决这一问题,提出一种基于纹理豪斯多夫距离(THD)的人脸识别算法。将人脸图像的空间量及纹理特征相结合,使其在深入的头部转动和人脸错位中都有很高的容错度。在FERET及Yale两大人脸数据库的实验表明,与其他经典的方法相比较,所提出的方法取得了更好的识别效果。     

网络结构增长的极端学习机算法

针对极端学习机(extreme learning machine,ELM)结构设计问题,基于隐含层激活函数及其导函数提出一种前向神经网络结构增长算法.首先以Sigmoid函数为例给出了一类基函数的派生特性:导函数可以由其原函数表示.其次,利用这种派生特性提出了ELM结构设计方法,该方法自动生成双隐含层前向神经网络,其第1隐含层的结点随机逐一生成.第2隐含层的输出由第1隐含层新添结点的激活函数及其导函数确定,输出层权值由最小二乘法分析获得.最后给出了所提算法收敛性及稳定性的理论证明.对非线性系统辨识及双螺旋分类问题的仿真结果证明了所提算法的有效性.     

磁盘阵列环境下的取证目标系统重构

封存原始硬盘与业务系统服务的连续性是计算机取证中面临的一对矛盾。在独立硬盘环境中,问题的解决并不困难,但在磁盘阵列(RAID)环境下,目前还缺乏有效的解决办法。为此,为RAID环境下的取证目标系统的重构提出一套解决方案,并设计实现其中关键任务的处理软件。实验表明,该方法可适用于不同RAID等级、不同操作系统下的系统重构,使得用户感受不到重构得到的系统与原系统有任何差异。与数据同步方法相比,该方法可以有效地减少重构的时间开销,且不会破坏原始证据硬盘上的数据。     

基于网络统考课程的学习支持服务构建

为满足学生在网络统考课程学习中自主学习的需要,学习支持服务系统的研究、设计构建和有效运行至关重要。各试点高校关于网络统考学习支持服务的实践和做法很多,但系统性研究和理论总结不够,作者结合工作实际尝试构建了网络统考学习支持服务体系基本框架。     

在文明寨地区的高精度地震资料处理实践

文明寨地区构造复杂，断块多，岩性变化大，一些地质问题不明，需要高精度三维资料处理来提高资料的纵、横向分辨率^[1][2]及为储层、油藏研究提供精确的依据。针对东濮文明寨地区高精度处理过程中的关键技术进行了大量分析试验，形成了一套适合东濮地区的较为完整的3D高精度地震资料处理流程。实际资料处理表明效果较好。