Efficient data selection for speech recognition based on prior confidence estimation using speech and monophone models

This paper proposes an efficient speech data selection technique that can identify those data that will be well recognized. Conventional confidence measure techniques can also identify well-recognized speech data. However, those techniques require a lot of computation time for speech recognition processing to estimate confidence scores. Speech data with low confidence should not go through the time-consuming recognition process since they will yield erroneous spoken documents that will eventually be rejected. The proposed technique can select the speech data that will be acceptable for speech recognition applications. It rapidly selects speech data with high prior confidence based on acoustic likelihood values and using only speech and monophone models. Experiments show that the proposed confidence estimation technique is over 50 times faster than the conventional posterior confidence measure while providing equivalent data selection performance for speech recognition and spoken document retrieval.     

含CO2气体混合物在60℃时的混合性质

测定了６０℃和２－６００ｂａｒ时由ＣＯ２和Ｈｅ、Ａｒ、ＳＦ６以及ＣＨＦ３组成的等分子二元混合物的压缩因子，根据纯物质和混合物的压缩因子计算了压力混合效应和过剩体积。对于真实气体混合性质的描述，本文采用五参数的ＢＡＣＫ方程，方程计算值与实验值符合较好。     

改进的GrabCut方法在舌诊系统中的应用

数字舌象图片的正确分割是实现计算机中医舌诊自动化系统的重要前提,为了高效而准确地分割出舌象,提出了一种基于四叉树与GrabCut的舌象分割方法.该方法首先利用四叉树分解对采集的舌象初分割,然后用相似区域的颜色均值优化GrabCut算法中高斯混合模型参数,最终完成舌象分割.实验结果表明：该算法使得舌象分割效率大大提高,具有很好的实用性.     

基于混合PSO的高斯混合模型地形分类

提出了一种基于改进的混合粒子群优化(particle swarm optimization,PSO)算法的高斯混合模型地形分类方法。高斯混合模型的求解通常是使用期望最大化算法(expectation maximization,EM),然而EM算法易陷入局部最优,收敛速度不稳定且对初值敏感。因此引入混合PSO算法,并对其进行了一系列改进。实验结果表明:改进后的算法较其它优化算法提高了全局搜索能力和收敛速度,利用该算法求解高斯混合模型可以提高参数估计的精度,并且在户外场景图像的地形分类实验中所提出的地形分类方法也表现优良。     

应用数字散斑技术的沥青混合料劈裂应变研究

采用数字散斑相关方法对沥青混合料试件间接拉伸试验中的拉伸应变进行研究.通过程序模拟散斑图的拉伸试验,得出所用数字散斑相关方法处理程序可用,且具有很高精度;通过与应变片测量法、计算法的比较和分析得出数字散斑相关方法的有效性.结果表明,数字散斑相关方法用于测量间接拉伸应变有效可行.     

冷再生结构层力学响应分析

通过研究冷再生结构层受力分析,在室内试验数据基础上,采用有限元BISAR软件对冷再生结构层以及常规半刚性路面结构层进行力学响应分析,得出了常规半刚性基层与冷再生基层力学响应的差异、冷再生层合理厚度、冷再生基层路面铣刨深度以及冷再生基层路面的使用条件.该研究对冷再生结构层设计与施工具有一定的借鉴作用.     

Pressureless low temperature sintering of nanocrystalline zirconia ceramics via dry powder processing

Pressureless sintering approaches provide a simple avenue to manufacture dense ceramic parts with minimal processing equipment, but current pressureless sintering techniques have yet to demonstrate capabilities of producing dense ceramics while maintaining sub-50 nm grain sizes. Nanocrystalline yttria stablized zirconia ceramics were process from 4 mol% yttria stablized zirconia (4YSZ) nanopowders with a crystallite size of 7.5 nm using dry cold isostatic pressing (CIP) where powders are dried immediately prior to green compact formation and CIP vacuum bagging. It is shown that CIP pressures >75 000 psi (517 MPa) effectively remove pores larger than 100 nm and that pressureless sintering occurs at reduced temperatures for green densities ≥50%. Though the sintering kinetics are shown to be similar to other zirconia nanopowder sintering studies, the small initial crystallize size and reduced sintering temperature allowed densities as high as 97.2%, while retaining a ceramic grain size at or below 40 nm. Produced nanocrystalline 4YSZ ceramics with a grain size of 30.3 nm and a density of 96.3% had Vicker's hardnesses as high as 14.2 GPa and Vicker's indentation fracture resistance of 3.43 MPa·, demonstrating that simple processing approaches can be refined to fabricate nanocrystalline ceramics while maintaining high hardness and indentation fracture resistance.     

Method of calculating the collision integral and solution of the Boltzmann kinetic equation for simple gases,gas mixtures and gases with rotational degrees of freedom

The conservative method of calculating the Boltzmann collision integral for simple gases, gas mixtures and gases with rotational degrees of freedom of molecules is presented. In all cases the common approach based on the projection technique of summing up the contributions in the collision integral is used. The method is applied for solving the Boltzmann kinetic equation for two fundamental rarefied gas flow problems: the heat transfer problem and the problem of shock wave structure. A comparison with experimental and numerical data of other authors is reported. It is shown that the considered numerical method allows one to solve the Boltzmann equation for real gases with high accuracy.     

空间约束混合高斯运动目标检测

目的 针对传统混合高斯模型前景检测运算量过大问题,提出一种基于空间约束的混合高斯前景检测算法。方法 通过快速初始化缩短模型的初始建立过程;采用双重背景模型机制,以自适应背景减法的前景检测结果作为混合高斯前景检测的空间约束条件,降低模型在背景区域的冗余运算;运用多策略自适应模型更新,提高前景检测的准确性。结果 在各种测试场景下,与传统混合高斯法、CodeBook、GMG、偏差均值混合高斯模型(MODGMM)等算法相比,该算法具有更好的准确率以及4倍以上的处理速度。结论 在固定相机场景下的运动目标检测中,算法能有效提高传统混合高斯法的准确性且具有极高的实时性。     

结合阴影抑制的混合高斯模型改进算法

混合高斯模型背景法作为运动目标检测的一种经典方法,已经广泛应用于智能视频监控系统中。但是,传统的混合高斯模型背景法容易将阴影误检测为运动目标的一部分。因此,针对该方法在区分阴影和运动目标方面的不足,提出了一种将混合高斯模型背景法和HSV空间阴影抑制相结合的运动目标检测算法。这种改进算法首先将颜色空间转换到HSV空间,初步提取运动目标,然后再利用阴影的灰度值比背景中的灰度值小,而前景的灰度值比背景中灰度值大的特性,检测出运动目标中的阴影。实验结果表明,这种改进的算法明显提高了检测效果,有效抑制了阴影对运动目标检测的干扰,算法实时性也较好。