A talking profile to distinguish identical twins

Identical twins pose a great challenge to face recognition due to high similarities in their appearances. Motivated by the psychological findings that facial motion contains identity signatures and the observation that twins may look alike but behave differently, we develop a talking profile to use the identity signatures in the facial motion to distinguish between identical twins. The talking profile for a subject is defined as a collection of multiple types of usual face motions from the video. Given two talking profiles, we compute the similarities of the same type of face motion in both profiles and then perform the classification based on those similarities. To compute the similarity of each type of face motion, we give higher weights to more abnormal motions which are assumed to carry more identity signature information.     

A coupled HMM approach to video-realistic speech animation

We propose a coupled hidden Markov model (CHMM) approach to video-realistic speech animation, which realizes realistic facial animations driven by speaker independent continuous speech. Different from hidden Markov model (HMM)-based animation approaches that use a single-state chain, we use CHMMs to explicitly model the subtle characteristics of audio-visual speech, e.g., the asynchrony, temporal dependency (synchrony), and different speech classes between the two modalities. We derive an expectation maximization (EM)-based A/V conversion algorithm for the CHMMs, which converts acoustic speech into decent facial animation parameters. We also present a video-realistic speech animation system. The system transforms the facial animation parameters to a mouth animation sequence, refines the animation with a performance refinement process, and finally stitches the animated mouth with a background facial sequence seamlessly. We have compared the animation performance of the CHMM with the HMMs, the multi-stream HMMs and the factorial HMMs both objectively and subjectively. Results show that the CHMMs achieve superior animation performance. The ph-vi-CHMM system, which adopts different state variables (phoneme states and viseme states) in the audio and visual modalities, performs the best. The proposed approach indicates that explicitly modelling audio-visual speech is promising for speech animation.     

HyperMask – projecting a talking head onto a real object

Published online: 15 March 2002     

基于Viseme的连续语音识别系统及Talking Head

为实现听觉/视觉驱动的说话人头部动画,该文给出了一个基于viseme(说话时的基本嘴形单位)的连续语音识别系统。它训练viseme隐马尔可夫模型(HMM),识别语音为viseme图像序列。建模采用triseme的概念来考虑viseme的上下文相关性,但它需要超大量的训练数据。该文根据viseme图像及其相似度权值(VSW)定义视觉问题集,用来建立triseme决策树,以实现triseme的状态捆绑及HMM参数共享。为比较系统性能,基于phoneme(听觉领域的语音基本单位)的语音识别结果也被映射为viseme序列。在评价准则上,定义viseme图像相似度加权识别精度,更全面地考虑输出和参考图像序列的差别,并用嘴形圆度和VSW曲线中的突变点来评估所得viseme序列的平滑性。结果表明,基于viseme的语音识别系统能给出更平滑和合理的嘴形图像序列。     

一种用于说话人头部动画的Viseme语音识别系统

Viseme是在语音驱动说话人头部动画中一种常用的为口形建立的音频-视频模型。本文尝试建立viseme隐马尔可夫模型(HMM)，用于驱动说话人头部的语音识别系统，称为前映射系统。为了得到更精确的模型以提高识别率，引入考虑发音口形上下文的Triseme模型。但是引入Triseme模型后，随着模型数量的急剧增加将导致训练数据的严重不足。本文使用决策树状态捆绑方法来缓解这一问题，同时引入了一种以口形相似度为基础的决策树视频问题设计方法。为了比较viseme系统的性能，本文也建立了一个以phoneme为基本HMM模型的语音识别系统。在评价准则上，使用了一种客观评价说话人头部动画的加权识别率。实验表明，以viseme为基本HMM模型的前映射系统可以为说话人头部提供更加合理的口形。     

User evaluation: Synthetic talking faces for interactive services

facial animation (FA) will be. We have undertaken experiments on 190 subjects in order to explore the benefits of FA. Part of the experiment was aimed at exploring the objective benefits, i.e., to see if FA can help users to perform certain tasks better. The other part of the experiment was aimed at subjective benefits. At the same time comparison of different FA techniques was undertaken. We present the experiment design and the results. The results show that FA aids users in understanding spoken text in noisy conditions; that it can effectively make waiting times more acceptable to the user; and that it makes services more attractive to the users, particularly when they compare directly the same service with or without the FA.     

《潜夫论》词语考释五则

考辨了东汉王符所著《潜夫论》中的 5条词语 ,主要针对汪继培笺和彭铎校正 ,提出了新的看法。     

关于中职学校计算机专业改革的探讨

21世纪的重要特征就是数字化、网络化与信息化,其基础是功能强大的计算机网络。计算机及其应用是发展最快的行业,也是国家紧缺型人才培养计划的重要方面。如何把学生培养成为高素质的国家实用性人才,这是中职学校必须面对的问题,也是必须在教育教学过程中落实的问题。     

Safety-critical event risk associated with cell phone tasks as measured in naturalistic driving studies: A systematic review and meta-analysis

A systematic review and meta-analysis of naturalistic driving studies involving estimates of safety-critical event risk associated with handheld device use while driving is described. Fifty-seven studies identified from targeted databases, journals and websites were reviewed in depth, and six were ultimately included. These six studies, published between 2006 and 2014, encompass seven sets of naturalistic driver data and describe original research that utilized naturalistic methods to assess the effects of distracting behaviors. Four studies involved non-commercial drivers of light vehicles and two studies involved commercial drivers of trucks and buses. Odds ratios quantifying safety-critical event (SCE) risk associated with talking, dialing, locating or answering, and texting or browsing were extracted. Stratified meta-analysis of pooled odds ratios was used to estimate SCE risk by distraction type; meta-regression was used to test for sources of heterogeneity. The results indicate that tasks that require drivers to take their eyes off the road, such as dialing, locating a phone and texting, increase SCE risk to a greater extent than tasks that do not require eyes off the road such as talking. Although talking on a handheld device did not increase SCE risk, further research is required to determine whether it indirectly influences SCE risk (e.g., by encouraging other cell phone activities). In addition, a number of study biases and quality issues of naturalistic driving studies are discussed.     

Inhalation of expiratory droplets in aircraft cabins

Airliner cabins have high occupant density and long exposure time, so the risk of airborne infection transmission could be high if one or more passengers are infected with an airborne infectious disease. The droplets exhaled by an infected passenger may contain infectious agents. This study developed a method to predict the amount of expiratory droplets inhaled by the passengers in an airliner cabin for any flight duration. The spatial and temporal distribution of expiratory droplets for the first 3 min after the exhalation from the index passenger was obtained using the computational fluid dynamics simulations. The perfectly mixed model was used for beyond 3 min after the exhalation. For multiple exhalations, the droplet concentration in a zone can be obtained by adding the droplet concentrations for all the exhalations until the current time with a time shift via the superposition method. These methods were used to determine the amount of droplets inhaled by the susceptible passengers over a 4-h flight under three common scenarios. The method, if coupled with information on the viability and the amount of infectious agent in the droplet, can aid in evaluating the infection risk. PRACTICAL IMPLICATIONS: The distribution of the infectious agents contained in the expiratory droplets of an infected occupant in an indoor environment is transient and non-uniform. The risk of infection can thus vary with time and space. The investigations developed methods to predict the spatial and temporal distribution of expiratory droplets, and the inhalation of these droplets in an aircraft cabin. The methods can be used in other indoor environments to assess the relative risk of infection in different zones, and suitable measures to control the spread of infection can be adopted. Appropriate treatment can be implemented for the zone identified as high-risk zones.