A segmentation scheme for head-shoulder video in MPEG compressed domain

More attention has been paid to the study of video object segmentation in compressed domain these years, which has already led to some practical technology. In this paper, a scheme is put forward for segmentation of head-shoulder video in MPEG （Motion Picture Experts Group） compressed domain. The conception of DCT （Discrete Cosine Transform） feature plane is defined. In the suggested scheme, firstly, the face region is detected by clustering skin-tone DCT feature points in the DCT feature plane. Secondly, the region of head-shoulder is approximately regarded as combination of the head rectangle and shoulder rectangle, and head rectangle is confirmed by double template matching. Thirdly, Canny operator and morphological operation are applied to the region of head-shoulder in feature plane to get the object mask and the region of object mask is rectified by correlation of DCT blocks to get high-quality segmentation.     

基于小波域TS-MRF模型的监督图像分割方法

定义在单一空间分辨率上的树结构马尔可夫场(Tree-Structured Markov Random Field,TS-MRF)模型能够表达图像的分层结构信息,但难以描述图像的非平稳性,针对该问题,提出小波域的TS-MRF图像建模方法-WTS-MRF模型,按照图像分类层次树的结构形式,该模型将一系列的MRF嵌套定义在多...     

基于文本分类技术的垃圾邮件识别系统

文章介绍了一个基于文本分类技术识别垃圾邮件系统的体系结构，并介绍了该系统涉及到的中文信息处理、文本特征选取、朴素贝叶斯分类器等关键技术。最后，文章给出了针对部分垃圾邮件的处理结果。结果表明，该方法对于垃圾邮件的识别，具有较好的效果。     

基于多帧边缘差异的视频运动对象的分割与跟踪算法

从视频场景中分割和跟踪感兴趣的视频对象对于MPEG-4等基于对象的视频编码来说是关键性的技术之一。针对目前大部分视频对象分割和追踪算法相当复杂但仍不能有效地去除背景噪声的问题,该文提出用于分割和跟踪视频运动对象的一种基于多帧边缘差异的算法。该算法利用一组帧的边缘差异来提取运动对象区域,通过聚类方法去除背景像素点,利用形态学算子得到对象分割模板,同时通过建立前帧感兴趣对象与当前帧运动对象的帧间向量跟踪当前帧的感兴趣视频对象。不同标准视频测试序列的测试结果表明,该算法能够实现对感兴趣的视频运动对象更为精确、快速和有效地分割和跟踪。     

一种两步判决的说话人分割算法

为了提高说话人分割(SS)准确率,该文综合考虑了静音信息和性别信息在SS中的作用,提出了一种两步判决的SS算法。在从音频流中分离出语音段的基础上,采用两步判决的方法进行SS。第1步采用基频信息为主、性别模型为辅的策略进行SS,将相邻说话人基频差异大的说话人改变检测出来;第2步采用基于性别的改进T2判决公式进行SS,实现相邻说话人基频差异小的同性别SS,为此,该文提出了一个基于块的潜在说话人改变点检测算法。实验结果表明,本文算法提高了分割准确率,F1度量值可达85.14%。对于短时长(2 s)语音段的SS,该算法和传统的贝叶斯信息判决算法相比,漏检率减少了16%。     

Internal preference mapping and the issue of satiety

Internal preference mapping (IPM) and Landscape Segmentation Analysis® (LSA) are two techniques broadly used to unfold consumers’ overall product liking ratings and create spatial maps that will provide further insights on consumers’ preferences. IPM is based on a vector model while LSA involves an ideal point model. Through a simulation and the analysis of 27 market research data sets, it is shown that IPM consistently creates a hedonic dimension that prevents the identification of satiety prone attributes (intensities higher or lower than a optimal level being disliked by the consumers) on that dimension. As a result, subsequent steps taken upon generating an IPM map such as the investigation of drivers of liking, population segmentation and the estimation of optimal product profiles have also a strong likelihood of resulting in distorted results, the level of distortion being dependent on the actual configuration of the underlying structure that IPM tried to uncover. It is also shown that a technique based on ideal points such as LSA does not exhibit this systematic artifact when unfolding liking data. Consequently, sensory scientists and market researchers should use caution when interpreting and using results issued from an internal preference mapping analysis.     

可变类谱聚类遥感影像分割

为实现遥感影像分割中类别数的准确、自动判别,提出了一种可变类谱聚类算法.根据影像的相似图构建权值矩阵和标准Laplacians矩阵,计算Laplacians矩阵较小特征值对应的特征向量生成特征向量矩阵,并视其与像素对应的向量行为像素特征点集;研究Laplacians矩阵处于不同（近似）块对角结构时类属同一目标类像素特征点的聚集性,定义聚类度指标,计算不同分割类别数对应聚类度;选择聚类度将发生最后一次较大跳变时的分割类别数作为算法估计类别数,并采用FCM（Fuzzy C-Means）算法划分该类别数对应像素特征点集实现影像分割.分别采用提出算法和基于特征间隙的算法分割合成及真实遥感影像.实验结果表明提出算法可准确地判别影像类别数.     

维吾尔语广播新闻连续语音敏感词检索系统

首先介绍语音信号来源于新疆人民广播电台维吾尔语新闻的敏感词语音语料库的建设。然后用该语料库进行基于HMM的模型训练。模型训练中详细介绍识别基元端点检测、特征提取、矢量量化、码本构建、HMM模型训练过程和结果。最后用该语料库和HMM训练模型对维吾尔语广播新闻连续语音信号进行敏感词检索,并对检索结果进行分析。     

一种改进的位平面复杂度分割算法

对复杂度直方图的不连续性测度的隐写分析能有效地察觉以位平面复杂度分割(BPCS)隐写的秘密信息的存在。为提高隐写算法的安全性,提出一种改进的BPCS算法,即利用部分可嵌入块作为嵌入信息的调节块,消除由于嵌入秘密信息而造成的复杂度直方图的剧烈变化,确保不连续性测度不会出现明显的峰值。实验结果表明,该算法能有效地抵抗基于复杂度直方图的统计分析,同时也保持了原算法嵌入容量大的特点。     

A Novel Ni/YSZ Anode Image Segmentation Method for Solid Oxide Fuel Cell Electrodes Microstructure

For the three‐phase identification of solid oxide cuel cell (SOFC) electrode, this paper presents a novel segmentation method based on gaussian mixture model (GMM) for YSZ/Ni anode optical microscopy (OM) images. A coarseness‐entropy adaptive factor is defined to incorporate the spatial information based on markov random field (MRF) into GMM. Furthermore, the proposed method can obtain the trade‐off between robustness to noise and effectiveness of preserving the details. Experimental results show that the proposed method outperforms the compared method on three‐phase microstructure identification. It can provide reliable foundation for the quantification of SOFC microstructure parameters.