改进的Faster RCNN煤矿井下行人检测算法

针对煤矿井下环境恶劣、光照差、背景混杂、行人模糊、行人多尺度等问题，提出了一种改进的Faster RCNN煤矿井下行人检测方法，使用深度卷积神经网络代替传统的手工设计特征方式自动地从图片中提取特征。利用深度学习通用目标检测框架Faster RCNN，以Faster RCNN算法为基础，对候选区域网络（Region Proposals Network，RPN）结构进行了改进，提出了一种“金字塔RPN”结构，来解决井下行人存在的多尺度问题；同时算法中加入了特征融合技术，将不同卷积层输出的特征图进行融合，增强煤矿井下模糊、遮挡和小目标行人的检测性能。实验结果表明：改进的Faster RCNN可以有效解决井下行人检测问题，在井下行人数据集上获得了90%的检测准确率，并在公测数据集VOC 07上对改进算法进行了验证。     

西安实施公共汽车专用道(路)的可行性初探

公共汽车专用道（路）是一种速度较快、运送能力较大、建设周期短、投资省、见效快的公共客运道路，也是一种从常规公共交通过渡到快速轨道交通的中间形式。特别在西安市还不具备大量投资修建快速轨道交通的经济能力之前，公共汽车专用道是解决西安市公共汽车运营速度低、车辆拥挤的好办法之一，西安市在道路及管理方面有其自己的优势，应该利用这种优势，尽早修建西安市的公共汽车专用道（路）。     

深度学习背景下视觉显著性物体检测综述

视觉显著性物体检测是对人类视觉和认知系统的模拟,而深度学习则是对人类大脑计算方式的模拟,将两者有机结合可以有效推动计算机视觉的发展。视觉显著性物体检测的任务是从图像中定位并提取具有明确轮廓的显著性物体实例。随着深度学习的发展,视觉显著性物体检测的精度和效率都得到巨大提升,但仍然面临改进主流算法性能、减少对像素级标注样本的依赖等主要挑战。针对上述挑战,本文从视觉显著性物体检测思想与深度学习方法融合策略的角度对相关论述进行分类总结。1）分析传统显著性物体检测方法带来的启示及其缺点,指出视觉显著性物体检测的核心思路为多层次特征的提取、融合与修整;2）从改进特征编码方式与信息传递结构、提升边缘定位精度、改善注意力机制、提升训练稳定性和控制噪声的角度对循环卷积神经网络、全卷积神经网络和生成对抗网络3种主流算法的性能提升进行分析,从优化弱监督样本处理模块的角度分析了减少对像素级标注样本依赖的方法;3）对协同显著性物体检测、多类别图像显著性物体检测以及未来的研究问题和方向进行介绍,并给出了可能的解决思路。     

Using Faster Gas Chromatography Analyses in Brewing Analytics

Dominant trends in analytical chemistry include miniaturization in sample preparation techniques and faster run times to provide high‐throughput screening, fast process monitoring and fast method development. This study focused on the application of narrow bore gas chromatographic capillary columns, 0.18 mm internal diameter, for brewing analyses. On these capillary columns, faster analyses could be performed compared to conventional GC capillary columns using 0.32 to 0.53 mm internal diameters. The robustness of the state‐of‐the‐art faster capillary gas chromatography, without compromising resolution, has been demonstrated with the analyses of beer flavour compounds such as lower and higher alcohols, esters and other volatile compounds such as acetaldehyde and dimethyl sulphide in beer. These methods were able to reduce sample run times by 60%.     

Geometric features of the flame propagation process for an SI engine having dual‐ignition system

Flame front surface area and enflamed volume (the volume enclosed with flame front) is theoretically analysed for a spark‐ignition engine, having cylindrical disc‐shaped combustion chamber with two spark plugs located axisymmetrically on cylinder head, between cylinder axis and cylinder wall. Spherical flame front assumption is used. A computer code is developed based on purely geometric consideration of the flame development process in combustion chamber, and is used to investigate the effects of variations of spark plugs' locations on geometric features of the flame front. A comparison has also been made with a spark‐ignition engine having one spark plug at the same location. Copyright © 2002 John Wiley & Sons, Ltd.     

基于减法聚类和快速紧密性函数的SF-FCM

首先结合减法聚类和模糊C-均值聚类各自的优点,运用减法聚类自适应地确定模糊C-均值聚类（FCM）的初始聚类数;然后,提出了改进的紧密性函数,以此改进用于确定FCM聚类结构的有效性函数.改进后的紧密性函数将对聚类结果贡献不大的数据予以剔除,使得算法适应能力更强,执行速度更快.实验结果表明,该快速紧密性函数是有效的,而且计算速度更快.     

基于修正CycleGAN的声呐图像库构建方法研究

声呐图像数据集获取困难,导致很多水下工作无法正常开展,如水下目标检测与跟踪、声呐图像的超分辨等,因此构建充足的声呐图像数据库成为很多水下研究工作的重要前提条件。受光学图像与合成孔径雷达（Synthetic Aperture Radar,SAR）图像转换研究工作的启发,提出了基于CycleGAN实现声呐图像库的构建,即利用光学图像合成声呐图像,实现光学到声呐的图像风格迁移。通过对CycleGAN网络损失函数的改进,提高了声呐图像的合成效果。通过与Pix2Pix等图像风格迁移网络进行比较的实验结果证明,修正后的CycleGAN网络具有更好的图像风格迁移效果。最后用合成的声呐图像训练Mask RCNN目标检测网络,并用真实的声呐图像进行测试,训练得到的模型能够成功地检测出真实声呐图像中对应的目标,进一步验证了利用光学图像构建声呐图像库的有效性。     

Coverless Image Steganography Based on Image Segmentation

To resist the risk of the stego-image being maliciously altered during transmission, we propose a coverless image steganography method based on image segmentation. Most existing coverless steganography methods are based on whole feature mapping, which has poor robustness when facing geometric attacks, because the contents in the image are easy to lost. To solve this problem, we use ResNet to extract semantic features, and segment the object areas from the image through Mask RCNN for information hiding. These selected object areas have ethical structural integrity and are not located in the visual center of the image, reducing the information loss of malicious attacks. Then, these object areas will be binarized to generate hash sequences for information mapping. In transmission, only a set of stego-images unrelated to the secret information are transmitted, so it can fundamentally resist steganalysis. At the same time, since both Mask RCNN and ResNet have excellent robustness, pre-training the model through supervised learning can achieve good performance. The robust hash algorithm can also resist attacks during transmission. Although image segmentation will reduce the capacity, multiple object areas can be extracted from an image to ensure the capacity to a certain extent. Experimental results show that compared with other coverless image steganography methods, our method is more robust when facing geometric attacks.     

快速包交换技术在CDMA基站中的应用

针对ＡＴＭ及帧中继在窄带实时业务方面的局限性,提出了一种结合二者优点的变长包硬件交换方案;采用长度在１２０个字节以内的变长帧结构,根据帧头的ＤＬＣＩ值由硬件完成分组的传递与交换,以有效地支持窄带时话音业务及宽带数据业务。     

结合广义自回归预训练语言模型与循环卷积神经网络的文本情感分析方法

传统的机器学习方法在对网络评论文本进行情感极性分类时,未能充分挖掘语义信息和关联信息,而已有的深度学习方法虽能提取语义信息和上下文信息,但该过程往往是单向的,在获取评论文本的深层语义信息过程中存在不足。针对以上问题,提出了一种结合广义自回归预训练语言模型（XLNet）与循环卷积神经网络（RCNN）的文本情感分析方法。首先,利用XLNet对文本进行特征表示,并通过引入片段级递归机制和相对位置信息编码,充分利用了评论文本的语境信息,从而有效提升了文本特征的表达能力;然后,利用RCNN对文本特征进行双向训练,并在更深层次上提取文本的上下文语义信息,从而提升了在情感分析任务中的综合性能。所提方法分别在三个公开数据集weibo-100k、waimai-10k和ChnSentiCorp上进行了实验,准确率分别达到了96.4%、91.8%和92.9%。实验结果证明了所提方法在情感分析任务中的有效性。