基于HMM-SVM的图像型火焰识别

针对单一的隐马尔科夫模型在图像型火灾探测中误报率偏高的问题,提出了隐马尔科夫模型和支持向量机相结合的图像型火焰识别算法。对捕获到的图像进行运动区域检测和颜色分析,提取疑似火焰区域,利用隐马尔科夫模型计算疑似区域与火焰模型的相似度,并输入到训练好的支持向量机进行二次识别。实验结果表明,与传统单一隐马尔科夫模型相比,该方法可以有效地降低误报率,提高火焰识别准确性。     

改进Hu矩的ATM机异常行为识别研究

提出一种基于改进Hu矩和隐马尔可夫模型相结合的ATM机异常行为识别方法。对ATM机前用户存(取)款行为的视频序列用改进Hu变换提取运动目标的行为特征,采用Baum-Welch算法对用户的正常行为进行训练,并建立隐马尔可夫模型;最后通过模型输出测试样本序列的概率来识别异常行为。采用Matlab对ATM机用户运动行为的模拟视频进行实验仿真,结果表明:该方法对ATM机前的用户行为具有较高的识别率。     

基于多时窗共空间模式的HMM运动想象脑电识别

运动想象脑电具有识别效果不佳及复杂时序信号建模困难的问题；提出一种基于多时窗共空间模式的隐马尔可夫模型运动想象脑电识别方法,首先将运动想象脑电划分为多个短时窗信号,然后使用共空间模式提取特征序列,以滤除脑电通道间的冗余信息,最后采用前向-后相算法与Viterbi算法求解隐马尔可夫模型并完成分类识别；将本文方法在公开运动想象脑电数据集上进行实验，得到77.17%的分类正确率，相较隐马尔可夫模型算法提升了5.74%，验证了所提方法的有效性。     

基于HMM监控视频的异常事件检测

针对智能监控系统中的行为分析与识别,将隐马尔可夫模型（Hidden Markov model,HMM）应用到智能视频监控系统的异常事件检测中。首先应用背景差法将运动目标提取出来。其次将运动目标的形状、颜色和帧间变化度等特征编码,生成特征向量。训 练时将特征向量送入HMM训练得到隐马尔可夫模型需要的参数[WTHX]A和B[WTBZ],检测时将特征向量送入HMM检测系统检测是否有异常事件发生。最后的实验结果表明,该方法能快速有效地检测监控视频中的异常事件的发生。     

基于加权HMM的车辆电源系统状态预测

针对车辆电源系统状态趋势问题,提出了一种加权隐马尔可夫模型的状态预测方法。通过建立电源系统的隐马尔可夫模型,利用加权预测思想对隐马尔可夫模型中隐状态序列进行预测,将最大概率隐状态利用观测概率密度计算出状态观测值。通过对电压调节脉宽信号的导通率进行预测,并与BP神经网络和自回归（AR）模型对相同序列的预测结果进行对比,结果表明该方法对系统的状态变化具有较好的预测能力。     

运动区域提取和闪频分析并行的火焰检测算法

在火焰检测中对火焰运动区域提取和闪烁特征分析大都分开进行,本文在提取运动区域的同时分析该区域的闪频特性,即将火焰的运动特征和闪烁特征同时提取。首先基于Ohta颜色空间找出图像中具有火焰颜色的疑似区域,其次根据视频图像某个位置在一段时间内变化的程度和次数是否都达到一定程度提取具有闪烁特性的运动区域,最后根据具有火焰颜色的连通区域是否包含这种运动区域,且颜色区域与运动区域的面积比例是否达到一定比值,来判断该连通区域是否为火焰。实验结果表明该方法在提取运动区域的同时能排除不具火焰闪烁特征的前景,且能在运动区域提取不完整的情况下保持较高的火焰检测率和较低的误检率。     

基于MCM的block-DCT域盲隐写分析算法

针对block-DCT域隐写,基于马尔可夫链模型（Markov Chain Model,MCM）,提出了一种新的盲隐写分析算法。该算法分析研究了DCT系数的分布特点和统计特征,采用隐马尔可夫树（Hiding Markov Tree Mode,HMT）模型,实现了载体图像的预测;基于MCM构建二阶马尔可夫经验转移矩模型,捕捉块内和块间DCT系数的相关性,提取经验转移矩阵对角元素作为特征向量,构建特征向量夹角作为检测秘密信息是否存在的依据。实验结果证明：该检测方法比普通的隐写算法具有较高的可靠性和较好的综合性能。     

利用均匀离散曲波域LCHMM的图像降噪算法

提出了一种在均匀离散曲波域中利用局部上下文隐马尔可夫模型进行建模的图像降噪算法。介绍均匀离散曲波变换的特点,分析其系数的统计分布规律,表明适合用隐马尔可夫模型对其进行建模。通过期望最大化训练获取模型的参数,利用参数得到降噪图像的系数估计。分别对光学图像和高分辨率的SAR图像进行了降噪实验,与小波域、轮廓波域的局部上下文隐马尔可夫模型等降噪方法进行比较,结果表明,提出的算法能够有效地去除噪声,具有较强的边缘保持能力。     

家庭服务机器人的人脸检测、跟踪与识别研究

本文介绍了一个用于家庭服务机器人完成人脸检测、跟踪、识别的双目视觉系统。该系统首先采用人脸肤色模型结合相似度来检测人脸;然后通过基于颜色信息的CAMSHIFT算法跟踪运动的人脸;最后利用嵌入式隐马尔可夫模型对人脸进行识别。实验结果表明该系统能自动地检测、跟踪、识别人脸,而且该系统具有较良好的实时性和鲁棒性。     

基于隐马尔可夫模型的动力型下肢假肢步态预识别

为使动力型假肢膝关节协调配合人体的运动,关键是对人体行走步态进行有效预识别.本文利用安装在假肢接受腔上的加速度传感器和安装在足底的压力传感器采集人体的运动信息,根据人体运动的规律性和重复性特点,通过将隐马尔可夫模型引入到所获得的运动信息中来分析并预识别人体的运动步态.实验表明,基于隐马尔可夫模型的动力型下肢假肢的步态预识别方法是有效并且准确的.     

Improved IMM Filter for Tracking a Highly Maneuvering Target with Mixed System Noises

We propose a method of improving tracking filter performance of a highly maneuvering target with mixed system noises in this paper. A case study of an off-road high speed moving target is considered. The system noises consist of white Gaussian noises generated from target motion models and additional colored noises arising from the effect of rough and uneven terrain profile. we design the colored noise first order discrete Markov dynamic system representing terrain conditions. Tracking is done by using an IMM filter with discrete white noise acceleration and horizontal coordinated turn models. The designed colored noise dynamic model is augmented with each of the motion models. We use Kalman filter for linear DWNA model while extended and unscented Kalman filters are used for nonlinear HCT model. A test scenario is setup and simulations are carried out. For filter performance comparison purposes, two more cases are considered i.e., systems with white noncorrelated system noises and the system correlated noise cases. Results show that the proposed method outperforms the traditional error treatment methods in terms of robustness, small mean square error, and acceptable computation load and data processing time.     

Learning,Generation and Recognition of Motions by Reference-Point-Dependent Probabilistic Models

This paper presents a novel method for learning object manipulation such as rotating an object or placing one object on another. In this method, motions are learned using reference-point-dependent probabilistic models, which can be used for the generation and recognition of motions. The method estimates (i) the reference point, (ii) the intrinsic coordinate system type, which is the type of coordinate system intrinsic to a motion, and (iii) the probabilistic model parameters of the motion that is considered in the intrinsic coordinate system. Motion trajectories are modeled by a hidden Markov model (HMM), and an HMM-based method using static and dynamic features is used for trajectory generation. The method was evaluated in physical experiments in terms of motion generation and recognition. In the experiments, users demonstrated the manipulation of puppets and toys so that the motions could be learned. A recognition accuracy of 90% was obtained for a test set of motions performed by three subjects. Furthermore, the results showed that appropriate motions were generated even if the object placement was changed.     

基于马尔可夫随机场与活动轮廓的运动目标分割

针对视频序列图像中的运动目标分割,提出了将马尔可夫随机场模型和活动轮廓模型相结合的运动目标分割算法。该算法首先利用马尔可夫随机场模型的运动检测算法,得到运动目标的初始模板。在此基础上提取出活动轮廓模型的初始轮廓点,然后构造活动轮廓模型的能量函数。用改进的贪婪算法求得能量函数最小值,提取出运动目标的精确轮廓,从而得到具有精确边缘的运动目标。实验结果表明该算法能有效地分割和提取出视频序列中的运动目标。     

Learning switching dynamic models for objects tracking

Many recent tracking algorithms rely on model learning methods. A promising approach consists of modeling the object motion with switching autoregressive models. This article is involved with parametric switching dynamical models governed by an hidden Markov Chain. The maximum likelihood estimation of the parameters of those models is described. The formulas of the EM algorithm are detailed. Moreover, the problem of choosing a good and parsimonious model with BIC criterion is considered. Emphasis is put on choosing a reasonable number of hidden states. Numerical experiments on both simulated and real data sets highlight the ability of this approach to describe properly object motions with sudden changes. The two applications on real data concern object and heart tracking.     

强光干扰下的火焰前景提取技术研究

火焰前景提取是视频型火灾检测算法的重要步骤,也是后续火焰特征识别算法的基础。针对现有火焰前景提取算法在强光干扰下或在背景与火焰颜色相近时无法正确提取火焰前景的问题,提出一种新的火焰前景提取算法。首先通过计算瞬时运动区域和火焰颜色区域来确定初级疑似火焰区域;然后对初级疑似区域和非疑似区域制定不同的背景建模策略来得到运动前景;最后由运动区域和高亮区域得到最终的火焰前景。与4种已有的火焰前景提取算法的对比实验表明,该算法在复杂背景下的火焰前景提取准确率为96.2%,远高于现有算法;能适应不同类型的复杂背景,并且满足实时性要求。     

Hybrid fire detection using hidden Markov model and luminance map

Recently, fire detection is a hot research topic. Although many detection methods have been proposed, there exist high false alarms because of the interference of fire-colored moving object in the complex environments. In this paper, a hybrid method is proposed. First, we get the set of candidate fire regions. Then these candidate fire regions are analyzed to exclude the fire-colored moving object. Our contributions are using the hidden Markov model (HMM) based on spatio-temporal feature and the variance of luminance map motivated by visual attention, and combining both for fire detection. The wrong detection can be reduced greatly. Experiment results show our proposed method has a good performance and it is robust to be used in complex environment compared with previous algorithms.     

多特征融合的视频火焰检测算法研究

基于视频的火焰检测算法为解决传统感烟感温火焰检测方法受环境制约的问题提供了一条新的路径。通常的视频火焰检测算法主要利用火焰的颜色、形状、频域特征等信息来进行检测,计算较为复杂,往往不能达到实时性。文中结合火焰的颜色、运动特性以及频闪特性,提出一种简单高效的视频火焰检测方法。首先使用ViBe算法提取出视频中的运动区域作为火焰候选区域,以降低计算量,再通过火焰的颜色模型筛选出疑似火焰区域,最后根据火焰的频闪特性建立一个简单的频闪模型,进一步滤除与火焰颜色相似的非火焰运动区域。通过实验证明,该文提出的算法能够检测出不同环境下火焰的发生,且执行效果较高。     

基于运动信息与马尔可夫彩色聚类的运动目标分割

针对视频序列图像中的运动目标分割,论文提出了将运动检测和马尔可夫彩色聚类相结合的运动目标分割算法。该算法首先利用基于统计模型的运动检测算法,通过后处理,得到运动目标的初始模板。然后,利用区域生长算法进行彩色图像的初始分割,在初始分割的基础上应用马尔可夫随机场模型进行彩色聚类,得到具有精确边缘的分割区域。最后,将运动目标的初始模板和彩色精确分割结合起来提取出具有精确边缘的运动目标。实验结果表明该算法能有效地分割和提取出视频序列中的运动目标。     

View-invariant motion trajectory-based activity classification and recognition

Motion trajectories provide rich spatio-temporal information about an object's activity. The trajectory information can be obtained using a tracking algorithm on data streams available from a range of devices including motion sensors, video cameras, haptic devices, etc. Developing view-invariant activity recognition algorithms based on this high dimensional cue is an extremely challenging task. This paper presents efficient activity recognition algorithms using novel view-invariant representation of trajectories. Towards this end, we derive two Affine-invariant representations for motion trajectories based on curvature scale space (CSS) and centroid distance function (CDF). The properties of these schemes facilitate the design of efficient recognition algorithms based on hidden Markov models (HMMs). In the CSS-based representation, maxima of curvature zero crossings at increasing levels of smoothness are extracted to mark the location and extent of concavities in the curvature. The sequences of these CSS maxima are then modeled by continuous density (HMMs). For the case of CDF, we first segment the trajectory into subtrajectories using CDF-based representation. These subtrajectories are then represented by their Principal Component Analysis (PCA) coefficients. The sequences of these PCA coefficients from subtrajectories are then modeled by continuous density hidden Markov models (HMMs). Different classes of object motions are modeled by one Continuous HMM per class where state PDFs are represented by GMMs. Experiments using a database of around 1750 complex trajectories (obtained from UCI-KDD data archives) subdivided into five different classes are reported.