基于Gabor变换的人眼定位

眼睛是人脸图像中最重要的特征点,有效地提取人眼特征在人脸识别中有重要的意义。鉴于2D Gabor函数与高等动物视觉皮层接受场的一致性,本文采用人眼的Gabor变换系数作为其特征模板,并用高斯多分辩搜索的方法,快速的定位人眼。实验结果表明,这种方法对于轻度旋转 、倾斜、面部表情的变化等有效好的鲁棒性。     

基于人眼视觉特性的小目标检测算法

人眼视觉特性一般应用于人眼观察数字图像的对比度增强领域,由于人眼视觉对连续数字图像中的小目标具有很强的检测能力,将人眼视觉特性引入小目标的检测领域,根据人眼视觉的侧抑制特性,以及在时间和空间上对变化的敏感特性,提出了基于人眼视觉特性的小目标检测算法.给出了一些仿真实验的结果,说明了此方法的有效性.     

基于人眼视觉的小波域结构相似性图像质量客观评价方法

针对空域结构相似性评价方法在几何失真图像及噪声污染图像的质量评价中存在的不足,在考虑人眼视觉特性前提下,研究了一种基于人眼视觉的小波域结构相似度的图像质量评价方法(WDSSIM),该方法首先将参考图像和失真图像进行小波变换,以获取不同尺度和频带的子带图像,并根据人眼视觉JND模型,获取人眼视觉对参考图像的带间敏感度和各子带的带内敏感度;然后,以带内敏感度因子为权值分别求取参考图像和失真图像同一尺度和频带的子带图像之间的结构相似度;最后,以带间敏感度因子为权值对各子带对的结构相似度进行加权归一化,获得整幅图像的结构相似度.     

基于人眼注视非穿戴自然人机交互

提出了一种基于人眼注视的非穿戴自然人机交互新方法。基于人类生物结构特征,采用主动形状模型确定人眼轮廓特征点,并根据HSV色彩空间中各色彩和亮度信息,构建综合反映人眼纹理及其空间位置的人眼特征直方图,采用粒子滤波方法,对人眼目标进行跟踪与定位,在克服光照和人脸姿态变化影响的同时,获取稳定的人眼轮廓特征点,满足人机交互的实效性要求。基于最大三角化划分人眼轮廓特征,构建人眼几何模型并确定人眼注视方向,通过图像帧间的均值滤波,确定人眼注视交互目标,用户无需佩带任何设备或有助于识别的辅助标志,实现非穿戴的自然人机交互。在用户灵活交互的同时,满足人机交互的舒适性和自由性要求。通过实验对比,验证了本文方法有效、可行。     

中值滤波算法在人眼视觉系统中的增强改进

本文提出一种在中值滤波算法基础上利用人眼视觉特性改进的图像去噪增强算法。该方法首先利用人眼特性将灰度级进行调整以提高图像背景亮度,然后在滤波模版中利用JND曲线特性提高像素间的对比度,从而在去除噪声的同时提高了图像的清晰程度,使人眼获得更多的信息量。实验结果表明:该方法与传统滤波算法相比,处理过的图像更利于人眼观察。     

基于人眼视觉特性的红外图像增强技术研究

基于红外图像灰度分布集中、对比度低的特点,结合人眼视觉特性,提出了基于人眼视觉特性的非线性变换方法.基于人眼灰度分辨能力随灰度变化的线性关系,推导了人眼的可分辨率灰度函数和基于人眼视觉的非线性变换函数,建立了人眼的非线性变换模型,实现了将有限的红外图像信息映射到有利于人眼观察的灰度分布区域.实验结果表明,该方法处理的图像反差适中、纹理明晰、细节部分对比明显,有效解决了红外图像低对比度和细节模糊的问题.     

基于嵌入式平台人眼疲劳检测终端的设计与实现

设计并实现一套基于Android平台的人眼疲劳检测终端。从摄像头中获取驾驶员的脸部图像数据,快速识别并定位到人脸部。再进行人眼部的定位,利用最大类间法Otsu对人眼进行目标提取以及Robert Cross边缘检测获得人眼的轮廓,从而进行人眼状态分析;利用PERCLOS原理进行疲劳的状态判断,并且借助开源视觉库Android-Open CV提高了系统的开发效率。在室内环境的实验结果表明该终端的表现良好。     

基于梯度积分投影和最大期望算法的人眼定位

针对传统积分投影方法易受眉毛、睫毛、阴影、遮挡及噪声等干扰的问题,提出了一种梯度积分投影与最大期望(EM)算法相结合的人眼精确定位方法,可以在人脸图像中分割出人眼区域,并精确定位人眼位置。首先,采用一种新的梯度算子计算人脸图像的行梯度积分投影粗略定位人眼区域;然后计算人眼区域的列梯度积分投影函数,用EM算法将所得列梯度积分投影函数曲线拟合成两个高斯曲线,并根据高斯曲线精确分割出人眼窗口;最后,利用我们提出的加权质心法在所得人眼窗口中精确定位双眼位置。在YaleB人脸数据库及自采数据库上的实验结果表明,本文方法不易受眉毛及噪声干扰,并能有效克服眼睑和睫毛的遮挡,对不同光照条件及头部姿态都有很好的鲁棒性。     

基于统计分析的水声低频线谱检测试验研究

针对浅海水声低频线谱检测应用场景，在理论仿真和海试数据处理基础上，采用统计分析的方法，对浅海低频环境噪声特性进行了分析和研究。对浅海低频环境下的线谱信号检测性能进行了数值模拟和仿真，获得了理论和试验环境下描述低频线谱信号检测性能的接收机工作特性（Receiver Operating Characteristic,ROC）曲线。理论和试验数据分析表明，浅海低频环境噪声基本符合理论描述的瑞利分布特征，线谱检测所需的信噪比判决门限略高于理论值，并随噪声的频率分布而变化。     

LED可变情报板的亮度调节

由于室外环境光的变化范围非常大,要使发光二极管(LED)可变情报板显示的信息在各种环境背景光下都清晰可见,就要求LED可变情报板的发光亮度能随环境背景光的变化作对应的调节。本文将从人眼的视觉特性、发光亮度与背景光的关系以及亮度调节的方法等方面讨论LED可变情报板的亮度调节,并给出测试结果。     

Identify autism spectrum disorder via dynamic filter and deep spatiotemporal feature extraction

Early intervention and treatment are crucial for individuals with autism spectrum disorder (ASD). However, it is challenging to identify individuals with ASD at an early age, i.e. under 3 years old, due to the lack of an effective and objective identification method. The mainstream clinical diagnosis relies on long-term observation of children’s behaviors, which is time-consuming and expensive, and thus how to accurately and quickly distinguish children with ASD in early childhood has become a critical issue. In this paper, we propose an eye movement based model to identify children with ASD. Specifically, children are required to freely observe some images. At the same time, their eye movements are recorded to analyze. Both the observed image and eye movements are input into our model. The input data are processed by the embedding layer, dynamic filters and LSTM block, respectively. Eventually, the spatiotemporal features are extracted to identify the eye movements belonging to a child with ASD or a typically developed child. Experiments on the Saliency4ASD dataset demonstrate that the proposed model achieves state-of-the-art performance in identifying children with ASD.     

Full Time-Varying Phase Noise Analysis for MOS Oscillators Based on Floquet and Sylvester Theorems

In this paper a set of full time-varying analyzing methods of phase noise for oscillators based on Floquet and Sylvester theorems are established, it provides a good idea for designing oscillators with perfect phase noise performance. The periodic state solution space of a linear periodic time-varying system is constructed with Floquet and Sylvester theorems, and the phase noise perturbation vectors of an oscillator autonomous system are characterized on this space. The analytical expressions of the phase noise spectrums, both 1/(Δ f)² and Lorentzian forms, are obtained, and the contributions to the phase noise of each noise sources are determined. With a generator approach and some modification, the method could be extended to the flicker noise. For RF front-end oscillators composed of MOS active devices, planar inductors and MOS varactors, the time-varying model parameters of the small signal equivalent circuits are constructed according to the periodic varying working-points. By the means of automatic small-signal equivalent-circuit construction, state-variable selection and periodic time-varying state-matrix generation, the system perturbation vectors and phase noise power spectrums are efficiently calculated. For a 10 GHz MOS oscillator, the 1/(Δ f)² and Lorentzian spectrums are calculated. Comparing with the results of SpectreRF, it indicates the proposed methods are accurate and reliable, especially the Lorentzian spectrum close to the carrier is more reasonable than previous methods. Every noise source contributions to the phase noise are listed and the results are analyzed. At last the applications of the methods to designing low phase noise oscillators and to analyzing the phase noise of composite systems, as well as the difficulty of flicker noise analysis, are addressed.     

基于NURBS向量肌肉模型的人脸动画及实现

当前,动画及其实现技术受到业界广泛关注,而人脸动画如喜、怒、哀、乐的表达其真实感还不够好.以Waters肌肉模型为基础,提出NURBS弹性肌肉模型,该方法依据解剖学知识,用非均匀有理B样条曲线仿真肌肉.通过改变曲线控制点的权重,可以找到一个动作向量控制肌肉的运动,进而合成人脸的各种表情.控制点数量越多,肌肉就越好控制,那么就可以更加真实地仿真人脸表情.     

A robust eye localization algorithm for face recognition

The accuracy of face alignment affects greatly the performance of a face recognition system. Since the face alignment is usually conducted using eye positions, the algorithm for accurate eye localization is essential for the accurate face recognition. In this paper, an algorithm is proposed for eye localization. First, the proper AdaBoost detection is adaptively trained to segment the region based on the special gray distribution in the region. After that, a fast radial symmetry operator is used to precisely locate the center of eyes. Experimental results show that the method can accurately locate the eyes, and it is robust to the variations of face poses, illuminations, expressions, and accessories.     

Short time Fourier analysis of the electromyogram: fast movements and constant contraction

We applied short-time Fourier analysis to surface electromyograms (EMG) recorded during rapid movements, and during isometric contractions at constant forces. We selected a portion of the data to be transformed by multiplying the signal by a Hamming window, then computed the discrete Fourier transform. Shifting the window along the data record, we computed a new spectrum each 10 ms. We displayed the transformed data in spectograms or "voiceprints." This short-time technique allowed us to see time-dependencies in the EMG that are normally averaged in the Fourier analysis of these signals. Spectra of EMG's during isometric contractions at constant force vary in the short (10-20 ms) term. Moments of the spectral distribution show this variability. Short-time spectra from EMG's recorded during rapid movements were much less variable. The windowing technique picked out the typical "three-burst pattern" in EMG's from both wrist and head movements. Spectra during the bursts were more consistent than those during isometric contractions. Furthermore, there was a consistent shift in spectral statistics in the course of the three bursts. Both the center frequency and the variance of the spectral energy distribution grew from the first burst to the second burst in the same muscle. We discuss this pattern with respect to the origin of the EMG bursts in rapid movement. We also extend the analogy between electromyograms and speech signals to argue for future applicability of short-time spectral analysis of EMG.     

The Three-Dimensional Spectrum Analysis of HDTV Luminance Signal

1　Ｔｈｅ 3 ＤＳｐｅｃｔｒｕｍＡｎａｌｙｓｉｓｏｆＨＤＴＶＬｕｍｉｎａｎｃｅＳｉｇｎａｌｓ　　Ｆｒｏｍｔｈｅｔｗｏ ｄｉｍｅｎｓｉｏｎａｌ (2 Ｄ )ｓｐｅｃｔｒｕｍａｎａｌｙｓｉｓｏｆＨＤＴＶｌｕｍｉｎａｎｃｅｓｉｇｎａｌｓ,ｔｈｅｓｔｉｌｌｉｍａｇｅｃａｎ     

液晶屏色彩谱线简易测量实验

为了了解液晶屏出射光的光谱,采用分光计和全息透射光栅设计了光谱测量的简易实验。通过实验测量了两部手机显示屏和一个电脑显示屏的显示光谱,结果表明,手机都包含红（R）绿（G）蓝（B）三个带状谱,手机液晶显示屏光谱与电脑液晶显示器相比,其RGB三色的谱宽大于电脑液晶显示器,色彩饱和度相对较低。文章所提简易方法可以用来粗略分析显示器显示色彩光谱。     

Electromagnetic scattering from vibrating penetrable objects using a general class of time-varying sheet boundary conditions

Calculation of electromagnetic (EM) scattering from vibrating penetrable cylinders of arbitrary cross-section is presented using a general class of time-varying sheet boundary conditions (SBCs) in conjunction with the method of moments (MoM). Sheet impedance and admittance expressions are first derived from the exact scattering solution for a penetrable circular cylinder with perturbed radius. Then, using the SBCs, integral equations are derived and solved numerically so that vibrating cylinders with arbitrary cross-section can be treated. Cylinder vibrations are assumed to be non-relativistic, allowing a simplified calculation of the scattered Doppler spectrum. A critical factor in the calculation of the potentially small Doppler components is that the time-varying nature of the cylinder boundary, contained within the sheet impedance and admittance expressions, can be isolated from the unperturbed terms in the scattered field. Comparison with exact and analytical perturbation solutions are presented to demonstrate the accuracy of the numerical solution.     

LTE系统中femtocell的动态频谱接入方法

针对LTE系统中femtocell的通信需求,给出一种基于认知无线电技术的动态频谱接入方案,提出一种上行频谱的感知方法,femtocell可发现周围被macrocell用户占用的频谱,还能进一步对这些频谱承载的业务类型进行识别,并根据不同的业务类型,选择不同的接入方法。如果femtocell检测到一段频谱承载的是话音业务,femtocell采用跳频的方法接入这段频谱;如果承载的是数据业务,femtocell就采用功率控制的方法接入频谱。仿真结果表明,该方法使femtocell实现对周围频谱承载业务的智能化识别和接入,从而在满足macrocell用户不同业务类型信干噪比要求的前提下,提高了femtocell小区的吞吐量。     

基于稀疏自动编码网络的水声通信信号调制识别

研究了基于稀疏自动编码网络的水声通信信号识别方法。首先利用稀疏自动编码网络对接收信号的功率谱识别分类,得到除PSK外信号的调制类型,然后对识别结果为PSK的信号做四次方谱,最后利用稀疏自动编码网络完成对QPSK和8PSK的识别分类。仿真实验表明,稀疏自动编码网络能从接收信号的谱信息中自动提取有效谱特征。与传统基于功率谱特征提取的识别方法相比,本文算法减少了依赖领域知识的特征提取环节,识别性能优于传统算法。