Flickr中的复合事件检测

Web资源上的复合事件检测已成为人们关注和研究的热点.Flickr作为一个照片共享的Web资源,其中的复合事件检测包括景点特色检测、用户兴趣检测等,从而主动地为用户提供个性化和智能化的服务.提出了一种新颖的、基于主题模型的方法来检测景点的旅游特色,以及同一景点在不同时节旅游特色的差异,并将其表示成一系列标签的概率组合.这种方法可以在没有用户标注的情况下进行自动检测,将非结构化的标签信息抽取成结构化的事件模式.Flickr中真实数据的实验结果表明该方法的可行性和有效性.     

集成组内标签与用户链接关系的Flickr组推荐模型

知名图片分享网站Flickr中的组,在增加图片曝光率、方便用户浏览图片等方面发挥重要作用.因而,如何协助用户便捷地搜寻其感兴趣的组成为研究者关注的热点之一.针对此问题,文中利用Flickr中辅助用户选择组的元素,即其关注的用户链接关系及组内的内容标签信息,提出一种组推荐模型.该模型基于概率矩阵分解的方法,具有较低的复杂度.在Flickr数据集上的实验表明,该模型可为用户提供较高质量的推荐结果.最后,基于该模型设计一个Flickr组推荐系统.     

基于空时极向LBP的极光序列事件检测

提出了一种用于检测全天空图像(ASI)序列中的弧状极光事件检测方法.针对弧状极光序列的运动趋势,在现有VLBP的基础上提出了基于空时极向LBP(ST-PVLBP)的极光序列事件检测算法,并用ST-PVLBP对极光序列进行表征.该算法结合序列帧间连续性信息和单帧空间位置信息,在保持高分类精度的同时降低了特征维数.在中国北极黄河站的ASI图像数据上的分类实验结果显示,所提出的方法可以有效检测全天空极光图像序列中的地磁南北向运动的弧状极光序列事件.     

一种面向主题的Flickr组内照片浏览方法

随着Web2。0技术的发展,以Flickr等为代表的社会标注类网站日益流行。Flickr组作为一种新兴的用户社区,不仅吸引大量在线用户的参与,也成为近年来研究者们关注的热点之一。本文针对Flickr里缺乏明确主题的组中的照片浏览的问题,提出一种面向主题的照片浏览方法。该方法由标签—照片的矩阵空间表示,基于非负矩阵分解算法的主题发现,以及综合考虑照片的浏览次数、好评次数和照片分享到组的时间信息建立的主题内照片影响力计算模型组成。本文在Flickr数据集上验证了该方法的有效性。这种基于主题的照片浏览方法能够有效地帮助用户了解和参与Flickr组。     

基于UGM灰预测模型的图像边缘提取算法

由于传统的边缘检测算法会产生信息漏检等问题,为了更有效检测出图像的边缘信息,提出了基于UGM灰预测模型边缘检测算法。该算法分别在垂直和水平方向上选取与考察点相邻的4个灰度值作为建模数据,通过UGM模型对建模数据进行处理,得到2幅预测图像,用原始图像分别减去2幅预测图像,根据正负得到4幅误差子图像,将4幅误差子图像相加,检测图像边缘。在此基础上,通过对误差子图像加入调节因子[q],提高边缘的清晰度。该算法与传统算法的结果比较表明该方法能清晰地检测出图像边缘点,图像的细节信息很好地保留下来,且对噪声图像具有一定的抗噪性,说明该算法是一种非常有效的图像边缘检测算法。     

面向图像自动语句标注的注意力反馈模型

图像自动语句标注利用计算机自动生成描述图像内容的语句,在服务机器人等领域有广泛应用.许多学者已经提出了一些基于注意力机制的算法,但是注意力分散问题以及由注意力分散引起的生成语句错乱问题还未得到较好解决.在传统注意力机制的基础上引入注意力反馈机制,利用关注信息的图像特征指导文本生成,同时借助生成文本中的关注信息进一步修正图像中的关注区域,该过程不断强化图像和文本中的关键信息匹配、优化生成的语句.针对常用数据集Flickr8k, Flickr30k和MSCOCO的实验结果表明,该模型在一定程度上解决了注意力分散和语句顺序错乱问题,比其他基于注意力机制方法标注的关注区域更加准确,生成语句更加通顺.     

基于局部能量的对称性检测算法

目前大多数对称性检测算法都是基于亮度或梯度信息进行的．文中分析了对称性与相位之间的关系,提出一种基于局部能量的对称性检测方法,将对称性检测问题转化为能量分析．在频域中用log Gabor小波对原始图像进行滤波,在不同尺度下局部能量最大、相位为特定相位且保持一致的点,即为物体的对称点．文中的可行性分析、算法定义及其合理性为该算法奠定了理论基础．实验表明,该算法可以克服目前已有算法需要图像分割的局限性,即该算法可直接应用于原始图像,不需要图像的任何先验知识,不需进行分割等任何预处理,而且还可以同时检测多种对称性．     

基于多特征提取的图像语义描述算法

针对图像语义描述方法中存在的图像特征信息提取不完全以及循环神经网络（RNN）产生的梯度消失问题,提出了一种基于多特征提取的图像语义描述算法。所构建模型由三个部分组成：卷积神经网络（CNN）用于图像特征提取,属性提取模型（ATT）用于图像属性提取,而双向长短时记忆（Bi-LSTM）网络用于单词预测。该模型通过提取图像属性信息来增强图像表示,从而精确描述图中事物,并且使用Bi-LSTM捕捉双向语义依赖,从而进行长期的视觉语言交互学习。首先,使用CNN和ATT分别提取图像全局特征与图像属性特征;其次,将两种特征信息输入到Bi-LSTM中生成能够反映图像内容的句子;最后,在Microsoft COCO Caption、Flickr8k和Flickr30k数据集上验证了所提出算法的有效性。实验结果表明,与m-RNN方法相比,所提出的算法在描述性能方面提高了6.8~11.6个百分点。所提算法能够有效地提高模型对图像的语义描述性能。     

传感器网络中基于环的负载平衡数据存储方法

传统的以数据为中心的存储方法有一个非常严重的缺点:网络中存在着明显的热点区域,主要出现在sink和home节点周围.位于热点区域附近的节点所消耗的能量远大于网络内的其他节点,从而严重缩短了传感器网络的寿命.对于出现在sink周围的热点问题,采用在网络中分布多个sink节点的方式消除热点.对于出现在home节点周围的热点问题,提出基于环的负载平衡数据存储协议,并以此数据存储为基础,给出相应的查询处理算法消除热点.首先,负载平衡数据存储协议将传感器网络划分为多个环,在网络工作的某个时间段内,数据被分散存储在某个环内的多个节点上;其次,该存储结构的最大特点是,在不同时间段内,各环轮换工作进一步消除热点.最后,基于环的查询处理算法也是由环内的多个节点协作完成.基于环的负载平衡数据存储协议以及基于此存储协议的查询处理算法虽然在查询处理中所消耗的能量高于传统方法,但可以保证从数据存储到查询处理的整个过程中,网络的所有节点均匀地消耗能量,从而避免了传统方法中的热点问题,达到延长网络寿命的目的.实验表明,基于环的数据存储及查询处理算法可以解决传统上基于事件存储协议中出现的热点问题,并延长传感器网络的寿命.     

基于社群隐含主题挖掘和多社群信息融合的自动图像标注

在Flickr图像共享网站上,大量无标签或者缺少标签的图像往往会因为标签信息的不完整,以致无法被有效地利用和检索。为了有效地进行图像检索,从Flickr用户经常会根据上传图像所隐含的主题而将其推荐到多个相关社群的特点出发,提出了一种新颖的基于社群隐含主题挖掘和多社群信息融合的自动图像标注算法。与传统的自动图像标注方法不同,该算法首先采用隐Dirichlet分配模型（latent Dirichlet allocation,LDA）对单个社群里的隐含主题（topic）进行挖掘,并利用隐含主题对由相似图像标签传播产生的初始“噪音”标签进行过滤;然后对同属于多个社群的图像,通过多社群信息融合来生成最终标注结果。实验结果显示了该新算法的有效性。     

基于LDA和word2vec的英文作文跑题检测

针对目前国内的英语作文辅助批阅系统缺少准确而高效的跑题检测算法的问题,提出了一种结合LDA和Word2vec的跑题检测算法。该算法利用LDA模型对文档建模并通过Word2vec对文档训练,利用得到的文档主题和词语之间的语义关系,对文档中各主题及其特征词计算概率加权和,最终通过设定合理阈值筛选出跑题作文。实验中通过改变文档的主题数而得到的不同F值,确定了最佳主题数。实验结果表明所提出的方法比基于向量空间模型的方法更具有效性,可以检测到更多的跑题作文并且准确率较高,F值达到89%以上,实现了作文跑题检测的智能化处理,可以有效地应用在英语作文教学中。     

An Aging Theory for Event Life-Cycle Modeling

An event can be described by a sequence of chronological documents from several information sources that together describe a story or happening. The goal of event detection and tracking is to automatically identify events and their associated documents during their life cycles. Conventional document clustering and classification techniques cannot effectively detect and track sequential events, as they ignore the temporal relationships among documents related to an event. The life cycle of an event is analogous to living beings. With abundant nourishment (i.e., related documents for the event), the life cycle is prolonged; conversely, an event or living fades away when nourishment is exhausted. Improper tracking algorithms often unnecessarily prolong or shorten the life cycle of detected events. In this paper, we propose an aging theory to model the life cycle of sequential events, which incorporates a traditional single-pass clustering algorithm to detect and track events. Our experiment results show that the proposed method achieves a better overall performance for both long-running and short-term events than previous approaches. Moreover, we find that the aging parameters of the aging schemes are profile dependent and that using proper profile-specific aging parameters improves the detection and tracking performance further     

基于轨迹分析的行人异常行为识别

提出一种基于轨迹分段主题模型的异常行为检测方法。为了解决跟踪偏差引起的轨迹不连续问题,首先使用模糊聚类算法对所有的轨迹进行全局聚类,然后对每一类轨迹采用分段采样的方式对段内轨迹点使用主题模型LDA进行局部聚类;以最大概率的轨迹点作为视觉单词,每类轨迹表示成一系列视觉单词的集合,在此基础上建立局部隐马尔科夫模型HMM;最后通过轨迹匹配的方法进行异常轨迹识别。在CAVIAR数据库上的实验结果表明,该算法能识别多种异常行为,提高了异常行为检测的准确率。     

A scalable algorithm for extraction and clustering of event-related pictures

The event detection problem, which is closely related to clustering, has gained a lot of attentions within event detection for textual documents. However, although image clustering is a problem that has been treated extensively in both Content-Based Image Retrieval (CBIR) and Text-Based Image Retrieval (TBIR) systems, event detection within image management is a relatively new area. Having this in mind, we propose a novel approach for event extraction and clustering of images, taking into account textual annotations, time and geographical positions. Our goal is to develop a clustering method based on the fact that an image may belong to an event cluster. Here, we stress the necessity of having an event clustering and cluster extraction algorithm that are both scalable and allow online applications. To achieve this, we extend a well-known clustering algorithm called Suffix Tree Clustering (STC), originally developed to cluster text documents using document snippets. The idea is that we consider an image along with its annotation as a document. Further, we extend it to also include time and geographical position so that we can capture the contextual information from each image during the clustering process. This has appeared to be particularly useful on images gathered from online photo-sharing applications such as Flickr. Hence, our STC-based approach is aimed at dealing with the challenges induced by capturing contextual information from Flickr images and extracting related events. We evaluate our algorithm using different annotated datasets mainly gathered from Flickr. As part of this evaluation we investigate the effects of using different parameters, such as time and space granularities, and compare these effects. In addition, we evaluate the performance of our algorithm with respect to mining events from image collections. Our experimental results clearly demonstrate the effectiveness of our STC-based algorithm in extracting and clustering events.     

LDA及主题词相关性的新事件检测

目前,话题检测与跟踪已被广泛应用,新事件检测作为话题检测与跟踪领域中的研究任务之一,为跟踪后续话题发展的先验知识,在话题检测与跟踪领域具有重要的理论研究意义。LDA主题模型不能自动识别新事件,其主题数需通过人工或反复实验来确定,识别效率低。本文提出基于LDA及主题词间的相关性新事件检测算法,同时结合报道发生的时间,确定合理的主题数目,从而探知新事件。实验证明,与传统LDA算法及Gibbs LDA算法相比,该方法具有一定优势,提高了对新事件检测的敏感度。     

基于主题模型的高分辨率遥感影像变化检测

提出一种基于主题模型的高分辨率遥感影像变化检测方法。将前后两期遥感影像对应的像素点对作为基本单位,提取其邻域亮度相关度、均值、标准差以及邻域回归直线的斜率、截距等低层次特征,在此基础上映射得到像素点对的高层次视觉单词特征,并通过潜在狄利克雷分配模型进行分析,挖掘其潜在的主题信息,即变化与不变,从而实现变化检测。实验结果表明,该方法能够有效检测高分辨率遥感影像的变化。     

Ranking canonical views for tourist attractions

Online photo collections have become truly gigantic. Photo sharing sites such as Flickr (http://www.flickr.com/) host billions of photographs, a large portion of which are contributed by tourists. In this paper, we leverage online photo collections to automatically rank canonical views for tourist attractions. Ideal canonical views for a tourist attraction should both be representative of the site and exhibit a diverse set of views (Kennedy and Naaman, International Conference on World Wide Web 297–306, 2008). In order to meet both goals, we rank canonical views in two stages. During the first stage, we use visual features to encode the content of photographs and infer the popularity of each photograph. During the second stage, we rank photographs using a suppression scheme to keep popular views top-ranked while demoting duplicate views. After a ranking is generated, canonical views at various granularities can be retrieved in real-time, which advances over previous work and is a promising feature for real applications. In order to scale canonical view ranking to gigantic online photo collections, we propose to leverage geo-tags (latitudes/longitudes of the location of the scene in the photographs) to speed up the basic algorithm. We preprocess the photo collection to extract subsets of photographs that are geographically clustered (or geo-clusters), and constrain the expensive visual processing within each geo-cluster. We test the algorithm on two large Flickr data sets of Rome and the Yosemite national park, and show promising results on canonical view ranking. For quantitative analysis, we adopt two medium data sets and conduct a subjective comparison with previous work. It shows that while both algorithms are able to produce canonical views of high quality, our algorithm has the advantage of responding in real-time to canonical view retrieval at various granularities.     

Automatic tag expansion using visual similarity for photo sharing websites

In this paper we present an automatic photo tag expansion method designed for photo sharing websites. The purpose of the method is to suggest tags that are relevant to the visual content of a given photo at upload time. Both textual and visual cues are used in the process of tag expansion. When a photo is to be uploaded, the system asks for a couple of initial tags from the user. The initial tags are used to retrieve relevant photos together with their tags. These photos are assumed to be potentially content related to the uploaded target photo. The tag sets of the relevant photos are used to form the candidate tag list, and visual similarities between the target photo and relevant photos are used to give weights to these candidate tags. Tags with the highest weights are suggested to the user. The method is applied on Flickr (). Results show that including visual information in the process of photo tagging increases accuracy with respect to text-based methods.     

基于统计主题模型的多粒度Web文档标注

针对已有Web文档语义标注技术在标注完整性方面的缺陷,将潜在狄里克雷分配(LDA)模型用于对Web文档添加语义标注。考虑到Web文档具有明显的领域特征,在传统的LDA模型中嵌入领域信息,提出Domain-enable LDA模型,提高了标注结果的完整性并避免了对词汇主题的强制分配;同时在文档隐含主题和文档所在领域本体概念间建立关联,利用本体概念表达的语义对隐含主题进行准确的解释,使文档的语义清晰化,为文档检索提供有效帮助。根据LDA模型可为每个词汇分配隐含主题的特征,提出多粒度语义标注的概念。在20news-group和WebKB数据集上的实验证明了Domain-enable LDA模型的有效性,并指出对文档进行多粒度标注有助于有效处理不同类型查询。