基于奇异值的具有年龄变化的人脸识别

为了增强现有人脸识别算法对年龄变化的鲁棒性,提出了一种新的基于奇异值分解(SVD)和嵌入式隐马尔可夫模型(EHMM)的人脸识别方法.先选取整幅人脸图像的奇异值作为基本特征向量,然后建立年龄函数,对奇异值特征进行修正,再根据得到的年龄函数,对人脸图像进行重建,提取改进后的奇异值特征作为观察序列,送入EHMM中进行分类识别,实验结果表明这种方法能够提高具有年龄变化的人脸识别效率.     

一种基于LBP-EHMM的人脸识别算法

针对实时人脸识别易受光照变化影响的问题,提出了一种将局部二值模式(Local Binary Patterns,LBP)与嵌入式隐马尔可夫模型(Embedded Hidden Markov Model,EHMM)相结合的人脸识别方法。该方法首先对输入的人脸图像进行LBP预处理,接着对其进行特征向量提取,最后把提取的特征观察向量送入EHMM进行训练或识别。在多个人脸数据库上进行了实验,结果表明该文算法对光照具有较好的鲁棒性,提高了识别率。     

基于混合EHMM模型的数据流预测

首先提出一种改进的算法NONEPI++,用于挖掘事件序列上非重叠发生的频繁情节;然后将每个频繁情节表示为相应的情节隐马尔可夫模型EHMM,并通过最大期望算法计算模型的混合系数,从而生成一个基于历史数据流的混合模型;最后,基于该混合模型预测目标事件类型出现的概率。实验表明,混合EHMM模型能有效地预测数据流。     

基于EHMM的人脸识别算法

人脸识别是对从视频图像中检测到的人脸区域进行身份的认证．是将待识别人脸与数据库中的人脸进行匹配的过程。将EHMM应用于人脸识别,提取人脸的DCT系数特征作为观察向量,用EHMM算法进行人脸模型训练和识别,并使用OpenCV对人脸识别算法进行功能仿真验证和相关探究,达到较好的人脸识别效果。实验结果表明,正常光照下,该算法的识别率在95％以上。     

虚拟人情感交互技术的研究

提出了基于EHMM的人工心理模型,该模型把情感感知和情感活动通过一个基本HMM过程有机地联系起来,研究结果表明：基于EHMM的人工心理模型符合人类情感变化规律。     

H.264JM模型中运动估计算法及改进方案

JM模型是JVT(joint video team)发布的H.264标准测试模型,对算法学习和研究有着重要的意义。根据JM测试模型的参数设定,其中的运动估计算法有3种可选模式。本文结合JM10.2的源代码对UMHexagonS算法进行了分析,并对该算法进行改进,能够在保证视频序列各分量信噪比的情况下缩短运动估计的耗时。本文利用UMHexagonS算法的准确预测以及运动估计代价的相关性来设置阈值达到提前结束搜索的目的。在JM10.2的测试模型上进行了算法验证。实验结果表明,利用块与块之间运动估计代价的相关性,在保证编码性能的同时,可以减少运动估计所需时间的10%以上。     

基于卷积与Transformer的人体姿态估计方法对比研究

人体姿态估计是计算机视觉的基础性算法之一,为了探究人体姿态估计领域的研究发展趋势,文章首先介绍了基于卷积的经典人体姿态估计算法,论述各算法的基本原理及算法改进,其次对最新的基于自注意力模型(Transformer)的算法进行梳理,最后介绍了常用的公开数据集和模型评价指标,选取了几个经典算法进行对比分析,平均精度在马克斯·普朗克信息研究所(Max Planck Institute Informatik,MPII)数据集达到80%以上,在微软公共对象上下文(Common Objects in Context,COCO)数据集达到60%以上,得到卷积结构和Transformer结构互有优劣的结论。     

基于离线模型预训练学习的改进DDPG算法

针对DDPG(deep deterministic policy gradient)在线训练过程中陷入局部极小值及产生大量试错动作和无效数据的问题,提出一种基于离线模型预训练学习的改进DDPG算法。利用已有数据离线训练对象状态模型和价值奖励模型,提前对DDPG中动作网络和价值网络进行预训练学习,减少DDPG前期工作量并提升在线学习的品质。加入DDQN(double deep Q-Learning network)结构解决Q值估计偏高问题。仿真结果中获取平均累积奖励值提升了9.15%,表明改进算法有效提高了DDPG算法效果。     

基于EHMM-SVM的人脸识别算法研究

EHMM依靠输出最大相似概率来判定人脸,但由于人脸图像的相似性,此方法可能会导致识别错误。对此,提出了一种基于EHMM-SVM的人脸识别方法。运用二维离散余弦变换(2D-DCT)进行人脸特征提取,得到观察向量序列。通过双重嵌套Viterbi算法求出每个人脸图像对应EHMM模型的输出概率,把输出概率输入SVM中进行分类训练以及识别测试,得到人脸识别的结果。运用ORL和YALE人脸数据库进行实验。实验结果表明了该方法的可行性及有效性。     

最小熵反馈式变结构多模型融合算法

统变结构多模型方法(VSMM)在处理高机动日标状态估计问题和大观测误差时存在因模型集合与真实模式匹配欠佳导致估计质量下降的问题.本文结合最小信息熵准则(ME)提出一种反馈式变结构多模型融合算法(MEVSMM),将在所有模型相关的在线估计信息进行反馈,进而选取状态估计分布信息熵最小的模型集作为当前有效模型集,计算多模型估计结果;结合粒子滤波算法(PF)和设计擂台赛算法(CM),构造了易于工程实现的次优算法(PF-MEVSMM).理论分析与仿真表明,与传统VSMM算法相比,本法具有模型集更精简、有效,融合估计结果鲁棒性更强、精度更高的优点.     

Facial expression recognition using ASM-based post-processing technique

Facial expression recognition is a challenging field in numerous researches, and impacts important applications in many areas such as human-computer interaction and data-driven animation, etc. Therefore, this paper proposes a facial expression recognition system using active shape model (ASM) landmark information and appearance-based classification algorithm, i.e., embedded hidden Markov model (EHMM). First, we use ASM landmark information for facial image normalization and weight factors of probability resulted from EHMM. The weight factor is calculated through investigating Kullback-Leibler (KL) divergence of best feature with high discrimination power. Next, we introduce the appearance-based recognition algorithm for classification of emotion states. Here, appearance-based recognition means the EHMM algorithm using two-dimensional discrete cosine transform (2D-DCT) feature vector. The performance evaluation of proposed method was performed with the CK facial expression database and the JAFFE database. As a result, the method using ASM information showed performance improvements of 6.5 and 2.5% compared to previous method using ASM-based face alignment for CK database and JAFFE database, respectively.     

Face recognition using the embedded HMM with second-order block-specific observations

The paper is concerned with face recognition using the embedded hidden Markov model (EHMM) with second-order block-specific observations. The proposed method partitions a face image into a 2-D lattice type, composed of many blocks. Each block is represented by the second-order block-specific observation that consists of a combination of first- and second-order feature vectors. The first-order (or second-order) feature vector is obtained by projecting the original (or residual) block image onto the first (or second) basis vector that is obtained block-specifically by applying the PCA to a set of original (or residual) block images. A sequence of feature vectors obtained from the top-to-bottom and the left-to-right scanned blocks are used as an observation sequence to train EHMM. The EHMM models the face image in a hierarchical manner as follows. Several super states are used to model the vertical facial features such as the forehead, eyes, nose, mouth, and chin, and several states in the super state are used to model the localized features in a vertical face feature. Recognition is performed by identifying the person of the model that provides the highest value of observation probability. Experimental results show that the proposed recognition method outperforms many existing methods, such as the second-order eigenface method, the EHMM with DCT observations, and the second-order eigenface method using a confidence factor in terms of average of the normalized modified retrieval rank and false identification rate.     

Fast retrieval of hidden data using enhanced hidden Markov model in video steganography

In the digital world, secure data communication has an important role in mass media and Internet technology. With the increase in modern malicious technologies, confidential data are exposed at a greater risk during data communication. For secured communication, recent technologies and the Internet have introduced steganography, a new way to hide data. Steganography is the growing practice of concealing data in multimedia files for secure data transfer. Nowadays, videos are more commonly chosen as cover media than other multimedia files because of the moving sequence of images and audio files. Despite its popularity, video steganography faces a significant challenge, which is a lack of a fast retrieval system of the hidden data. This study proposes a novel video steganography technique in which an enhanced hidden Markov model (EHMM) is employed to improve the speed of retrieving hidden data. EHMM mathematical formulations are used to enhance the speed of embedding and extracting secret data. The data embedding and retrieving operations were performed using the conditional states and the state transition dynamics between the video frames. The proposed EHMM is extensively evaluated using three benchmark functions, and experimental evaluations are conducted to test the speed of data retrieval using differently sized cover-videos. Results indicate that the proposed EHMM yields better results by reducing the data hiding time by 3–50%, improving the data retrieval rate by 22–77% with a minimum computational cost of 20–91%, and improving the security by 4–77% compared with state-of-the-art methods.     

具有不同数目状态结点的HMMs在中国手语识别中的应用

中国手语是中国聋人使用的语言,主要通过手势动作来表达一定的含义。因而,手语识别问题是动态连续信号的识别问题。目前大部分手语识别系统采用HMMs(hidden Markov models)作为系统的识别系统。由于各个词包含的基本手势数不同,若所有模型都由同样数目的状态结点构成会影响识别率。而由人为每个词设置状态数又很难达到完全准确,所述系统使用一种基于动态规划的估计状态结点数的办法,并实现了基于具有不同状态数目的HMM的训练及识别过程,实验结果表明,该系统在手语的识别速度和识别精度方面都有所提高。     

置信度判别嵌入式隐马尔可夫模型人脸识别

为了提高人脸识别率,提出了一种优化置信度的判别嵌入式隐马尔可夫（EHMM）人脸识别方法。提出的方法基于假设检验,通过最小化检验错误率得到优化置信度判别式训练准则。在优化置信度判别式训练准则的前提下,通过参数估计求解判别式转换矩阵,提取出具有判别性、低维度的图像特征,确保观察样本能正确地分配到其对应的模型状态,以提高所训练出的EHMM模型的正确识别率。理论分析证明了优化置信度判别式训练准则的有效性,详细的实验及与现有方法的比较结果表明,提出的识别方法具有更好的识别性能。     

A real-time face detection and recognition system for a mobile robot in a complex background

This article presents a real-time face detection and recognition system for mobile robots based on videos with a complex background. In the visual system, we propose a multi-information method consisting of an Adaboost algorithm, and color information for the face detection part. The interesting targets in the video will first be detected by the Adaboost algorithm, which is robust to illumination. Then the skin color model in YCbCr space will be employed to select the parts that may not be skin areas from the information detected by the Adaboost algorithm. An embedded hidden Markov model (EHMM) is presented, using a 2-DCT feature vector as the observation vector, to recognize the faces detected. The whole process of detecting and recognizing a frame, which is 320 × 240, will take 1.4 s with the rapid recognition parameters and 4.2 s with the slow recognition parameters.     

混合语音识别系统的一种新的简化神经网络结构

研究适用于隐马尔可夫模型（HMM）结合多层感知器（MLP）的小词汇量混合语音识别系统的一种简化神经网络结构。利用小词汇量混合语音识别系统中的HMM状态所形成的规则的二维阵列，对状态观测概率进行分解。基于这种利用HMM的二维结构特性的方法，实现了用一种由多个简单的MLP所组成的简化神经网络结构来估计状态观测概率。理论分析和语音识别实验的结果都表明，这种简化神经网络结构在性能上优于Franco等人提出的简化神经网络结构。     

EHMM人眼状态识别算法中的2D-DCT IP核设计

EHMM人眼状态识别算法具有较高的识别率但算法复杂,因此利用SOPC开发平台设计了一种基于行列变换快速算法的2D-DCT IP核以提高处理速度;根据眼状态识别只需取2D-DCT后左上角部分子矩阵数据的特点,对其计算过程进一步优化;并引入了转置存储技术,浮点数乘法通过移位后转化为定点乘法器实现,优化了硬件资源,提高了处理速度;实验结果表明该IP核很好地实现了人眼图像灰度值的DCT变换。