A Bayesian approach to vectorization of object boundaries from digital images and to geometrical uncertainty assessment

A new Bayesian approach is presented for extracting 2D object boundaries with measures of uncertainty. The boundaries are described by minimal closed sequences of segments and arcs, called mixed polygons. The sequence is minimal in the sense that it is able to describe all the geometrical properties of the boundary without being redundant. Based on geometrical measures evaluated on the object boundary model, a prior distribution is introduced in order to favor a mixed polygon with good geometrical properties, avoiding short sides, collinearity between segments, and so on. The estimation process is based on a two‐stage procedure that combines reversible‐jump MCMC (RJMCMC) and classic MCMC methods. The RJMCMC method is viewed as a model selection technique, and it is used to estimate the correct number of sides of the mixed polygon. The MCMC algorithm provides a sample of mixed polygons through which to evaluate the mixed polygon that best approximates the object boundary and its geometrical uncertainty. A convergence criterion for the RJMCMC method is provided. Copyright © 2011 John Wiley & Sons, Ltd.     

基于听感知特征的语种识别

为了在语种识别时充分利用人的听感知特性提高识别性能,提出了一种基于听感知模型的特征。听感知特征采用Gammatone滤波器组代替常用的三角滤波器组计算语音信号各子带能量;根据等效矩形带宽模型,确定各滤波器的中心频率与带宽;使用反置等响度曲线模拟人耳对信号不同频率成分的主观响度感受。在基本听感知特征的基础上,还提出了一、二阶差分特征和偏移差分特征用于语种识别。对比实验表明,该文所提的听感知特征性能均优于目前普遍使用的Mel频率倒谱系数(MFCC)特征及其衍生特征。     

基于螺栓接头受力性能的盾尾注浆压力控制

对采用螺栓接头的管片,管片间接头主要靠两种效应,即端面摩擦效应及螺栓抗剪效应,来平衡注浆压力对管片接头的剪切,认为端面摩擦效应主要在管片发生相对位移前发生作用,之后则螺栓抗剪效应是主要的.通过引入等效空隙率替代土体本身的空隙率来考虑建筑间隙的影响,对盾尾注浆情况下因注浆而对管片造成的压力进行了理论推导,得到注浆对管片产生的压力计算式.结果表明:注浆对管片产生的压力,随注浆时间的增长、注浆压力的增大而增大.将接头螺栓的抗剪效应与注浆对管片产生的压力结合起来,据接头螺栓的剪切破坏条件来控制注浆压力和注浆时间,即在求得注浆压力产生的螺栓剪应力前提下,要求该剪应力小于螺栓的许用剪应力,从而实现对注浆压力及注浆时间的控制.     

头相关传输函数与虚拟听觉重放

头相关传输函数（HRTF）是自由场情况下点声源到双耳的声学传输函数,包含有声源定位的主要物理信息,在双耳听觉物理的研究中是非常重要的.虚拟听觉重放是HRTF的一个重要应用,它采用HRTF信号处理的方法模拟声源到双耳声学传输过程,从而在声重放中产生相应的空间听觉事件.目前HRTF与虚拟听觉重放已成为物理学（声学）、信号处理、听觉生理等研究领域的热门与前沿课题,受到国内外多学科研究工作者的共同关注,并在众多的领域得到广泛的应用.该文综述HRTF与虚拟听觉重放的基本原理与国内外研究进展,并概述了虚拟听觉重放的一些重要应用.     

Corticofugal modulation of frequency tuning of inferior collicular neurons in big brown bat,Eptesicus fuscus

In order to explore the possible mechanism of corticofugal modulation of excitatory frequency tuning curves (EFTCs) of midbrain neurons, we examined the change of sharpness, frequency-intensity response area, minimum threshold of both EFTCs and inhibitory frequency tuning curves (IFTCs) of inferior collicular neurons during corticofugal modulation using two-tone inhibition paradigm and micro-electrical stimulation technique. Our data showed that corticofugal inhibition increased sharpness, minimum threshold, and decreased the frequency-intensity response area of EFTCs, at the same time it decreased the sharpness, minimum threshold but increased the frequency-intensity response area of IFTCs. The opposite results were observed for EFTCs and IFTCs of corticofugally facilitated inferior collicular neurons. During corticofugal inhibition, the percent change of frequency-intensity response area of EFTCs had significant correlation with the percent change of that of IFTCs. These data suggest that cortical neurons are likely to improve frequency information processing of inferior collicular neurons by modulation of IFTCs.     

Roles of computational auditory model in automatic speech recognition

The development of computational auditory modeling in the last 30 vears is reviewed, and its development in the com trig years is predicted .     

一种采用振荡神经网络的CASA计算模型语音分离算法

基于听觉现象分析（CASA）模型的基本原理,在仅有单通道输入混合语音信号时,采用振荡器神经网络,提出了一种CASA改进模型语音分离算法结构,文中利用一个实例说明了新算法的具体实现步骤,讨论了新算法机构中语音听觉外围处理部分和分割神经网络处理部分,通过上述两个部分的处理可以将输入混合语音信号在时频域上分割为若干有听觉感知意义的语音听觉感知成分分段Segments,以便于新算法后续处理分部中语音Segments的聚为和分离重构输出处理,最终完成语音分离任务。     

鼠耳蝠下丘听神经元对超声刺激的反应特性

实验在５只鼠耳蜗上进行，共观察了２２７个对超声刺激发生反应的下丘听神经元。这些神经元听反应的最佳频率在１４．３－７６．２ｋＨｚ之间；潜伏期在３．０－１２．０毫秒之间，多数神经元（７９．３％）为５．０－７．０毫秒；最低阈值位于－２０．０－７０．０ｄＢ ＳＰＬ之间；多数神经元（８２．０５％）的调谐曲线为宽阔型，少数（１７．９５％）为狭窄型；反应最佳频率沿下丘背腹轴呈明显有序地排列。     

基于代数和三角多项式加权的二次混合样条曲线

利用代数和三角多项式加权的方法,构造了一种二次混合样条曲线,这种曲线具有二次非均匀B样条曲线相似的性质.这里的权系数也是形状参数,称之为权参数,取值范围从[0,1]扩大到[-3.659 79,5.278 98].权参数的不同取值可以整体或局部地调整曲线的形状,并且权参数能像开关那样,使得曲线的各段非常方便灵活地在代数多项式、三角多项式之间转换.不需要用重节点或解方程组方法,而只要令某个或某些权参数取-3.659 79,曲线就能接插值于控制点或控制边.     

一种分层阈值优化的语音感知小波去噪方法

通过分析含噪语音信号的特点,引入能够兼顾人耳听觉特性的听觉感知小波变换,构造了新的小波阈值函数,并对小波变换分解后的阈值进行基于微粒群算法的分层优化.仿真实验表明,该方法在不同信噪比条件下均具有较好的去噪性能,语音的可懂度和听觉效果得到有效提高.