T型发射区单晶硅太阳电池的输出特性研究

利用TCAD半导体器件仿真软件对具有T型发射区结构的单晶硅太阳电池进行了仿真研究。全面系统地分析了在不同衬底少子寿命情况下,不同T型发射区深度对太阳电池外量子效率、短路电流密度、开路电压、填充因子及转换效率的影响。仿真结果表明:采用T型发射区结构可在一定程度上提高常规均匀发射区太阳电池的电学性能;T型发射区结构对700~1200nm长波段入射光的外量子效率具有明显的改善作用;当衬底少子寿命一定时,太阳电池短路电流密度、填充因子均随T型发射区深度的增大而增大,而开路电压随T型发射区深度的增大而减小;当T型发射区深度大于80μm时,对于低衬底少子寿命的单晶硅太阳电池,T型发射区结构对其转换效率的改善效果最为显著。     

Deep level transient spectroscopy on charge traps in high-k ZrO2

The deep trap properties of high-dielectric-constant (k) ZrO₂ thin films were examined by deep level transient spectroscopy (DLTS). The hole traps of a ZrO₂ dielectric deposited by sputtering were investigated in a MOS structure over the temperature range, 375 K-525 K. The potential depth, cross section and concentration of hole traps were estimated to be ∼ 2.5 eV, ∼ 1.8 × 10^− 16 cm² and ∼ 1.0 × 10¹⁶ cm^− 3, respectively. DLTS of ZrO₂ dielectrics can be used to examine the threshold voltage shift (?V_th) during the operation of SONOS-type flash memory devices, which employ high-k materials.     

Separation of Transparent Layers using Focus

Consider situations where the depth at each point in the scene is multi-valued, due to the presence of a virtual image semi-reflected by a transparent surface. The semi-reflected image is linearly superimposed on the image of an object that is behind the transparent surface. A novel approach is proposed for the separation of the superimposed layers. Focusing on either of the layers yields initial separation, but crosstalk remains. The separation is enhanced by mutual blurring of the perturbing components in the images. However, this blurring requires the estimation of the defocus blur kernels. We thus propose a method for self calibration of the blur kernels, given the raw images. The kernels are sought to minimize the mutual information of the recovered layers. Autofocusing and depth estimation in the presence of semi-reflections are also considered. Experimental results are presented.     

二元树复小波域的局部高斯混合模型图像降噪

在复小波域上对观测图像进行一种基于高斯混合模型的后验概率分类,并在每类小波系数的局部邻域估计出局部高斯混合模型的参数,这种参数估计是局部自适应的;然后用该局部高斯混合模型对各个子带系数进行贝叶斯框架下的最大后验概率(MAP)估计,以达到降低噪声的目的．由于这种小波变换具有近似的平移不变性和良好的方向选择性,因此在降噪的同时可以很好地消除主要边缘处的“震铃”效应．实验结果表明;文中算法无论从峰值信噪比还是从主观视觉效果上都要优于一些传统的降噪算法．     

An improved direct labeling method for the max–flow min–cut computation in large hypergraphs and applications

Algorithms described so far to solve the maximum flow problem on hypergraphs first necessitate the transformation of these hypergraphs into directed graphs. The resulting maximum flow problem is then solved by standard algorithms. This paper describes a new method that solves the maximum flow problem directly on hypergraphs, leading to both reduced run time and lower memory requirements. We compare our approach with a state–of–the–art algorithm that uses a transformation of the hypergraph into a directed graph and an augmenting path algorithm to compute the maximum flow on this directed graph: the run–time complexity as well as the memory space complexity are reduced by a constant factor. Experimental results on large hypergraphs from VLSI applications show that the run time is reduced, on average, by a factor approximately 2, while memory occupation is reduced, on average, by a factor of 10. This improvement is particularly interesting for very large instances, to be solved in practical applications.     

超分辨率重建技术及其在太赫兹图像中的应用

超分辨率重建就是通过相应的算法,重建图像截止频率之外的细节信息,重构出一幅清晰的高分辨率图像。首先介绍了超分辨率重建算法——非均匀内差法,迭代反投影法（IBP）,凸集投影法（POCS）,说明了各算法的概念和应用,并着重介绍了基于最大后验概率（MAP）的图像超分辨率算法,给出了MAP超分辨率复原算法处理实际太赫兹图像的结果。实验表明,超分辨率图像重建具有重建效果好、抗噪声性能强的优点,有效地重建了高分辨率太赫兹图像,在太赫兹成像领域具有良好发展和应用前景。     

基于场景模态深度理解网络的单目图像深度理解

基于深度卷积神经网络的图像处理方法得到的单目深度图像质量远高于传统图像处理方法,但该方法对无用特征的训练易产生误差积累,且基于回归求解的连续深度距离预测精度较低,导致图像深度信息提取不精确、目标边缘模糊与图像细节缺失.提出一种应用于单目彩色图像的场景模态深度理解网络.建立以堆叠沙漏为主框架的网络模型,通过反复进行自下而...     

基于深度语义模型的乳腺X线图像检索

图像特征是基于内容的图像检索（Content-based image retrieval,CBIR）的关键,大部分使用的手工特征难以有效地表示乳腺肿块的特征,底层特征与高层语义之间存在语义鸿沟。为了提高CBIR的检索性能,本文采用深度学习来提取图像的高层语义特征。由于乳腺X线图像的深度卷积特征在空间和特征维度上存在一定的冗余和噪声,本文在词汇树和倒排文件的基础上,对深度特征的空间和语义进行优化,构建了两种不同的深度语义树。为了充分发挥深度卷积特征的识别能力,根据乳腺图像深度特征的局部特性对树节点的权重进行细化,提出了两种节点加权方法,得到了更好的检索结果。本文从乳腺X线图像数据库（Digital database for screening mammography, DDSM）中提取了2 200个感兴趣区域（Region of interest,ROIs）作为数据集,实验结果表明,该方法能够有效提高感兴趣肿块区域的检索精度和分类准确率,并且具有良好的可扩展性。     

多特征组合的深度图像分割算法

深度图像直接反映景物表面的三维几何信息,且不受光照、阴影等因素的影响,对深度图像处理、识别、理解是目前计算机视觉领域研究的热点和重点之一。针对深度图像信息单一且噪声较大的特点,提出一种基于组合特征的阈值分割算法,实现对深度图像数据的有效分割。算法首先通过梯度特征对图像进行Otsu阈值分割;在此基础上,分别在不同分割区域内利用深度特征进行Otsu多阈值分割,得到候选目标;然后,在空域上利用像素的位置特征对候选目标进行分割、合并与去噪,最终得到图像分割的结果。实验结果表明,该方法能有效克服深度图像中噪声的影响,得到的分割区域边界准确,分割质量较高,为以后的室内对象识别和场景理解工作奠定了较好的基础。