GAT0 推进剂智能化检选系统

为解决人工检选GAT0 推进剂药管尺寸与外观质量存在的问题,对GAT0 推进剂智能化检选系统进行设 计。提出图像检测技术的自动化智能检选方案,采用基于图像检测方式检测药管尺寸和外观缺陷,利用专用仿真和 分析工具对系统进行分析,对智能检选系统的数据采集进行统计分析,结合推进剂机器视觉检选技术,得出智能检 选系统的实施效果。应用结果表明：该设计可提高产品质量和保障产品安全高效生产,有一定的实用参考价值。     

Flow boiling data and prediction method for enhanced boiling tubes and tube bundles with R-134a and R-236fa including a comparison with falling film evaporation

New enhanced boiling tubes from Wolverine Tube Inc. (Turbo-B5) and Wieland-Werke AG (Gewa-B5) were investigated using R-134a and R-236fa as test fluids. The tests were done at saturation temperatures of 5 and 15 °C, mass flow rates from 4 to 35 kg m⁻² s⁻¹ and heat fluxes from 15 to 70 kW m⁻². A new prediction method based on a theoretical analysis of thin film evaporation was used to propose a new correlating parameter. A large new database of local heat transfer coefficients was obtained and utilized to generate an improved prediction method for bundle boiling and the onset of dryout. Onset of dryout and the simultaneous reduction in heat transfer performance occurred at very high vapour quality on these enhanced tubes in convective bundle boiling. Furthermore, a direct comparison was made between the tubes operating in falling film and convective bundle boiling modes.     

聚晶金刚石复合片表面裂纹视觉检测技术研究

采用计算机视觉检测技术提取出表面缺陷特征量,完成聚晶金刚石复合片表面裂纹缺陷检测。首先,根据聚晶金刚石复合片表面特性,研究合适的光源照明系统。然后,提出一种基于直方图投影梯度极值的局部边界提取方法,将感兴趣区域进行提取。在此基础上,采用图像滤波、阈值分割的方法实现裂纹的准确提取。最后,通过计算裂纹连通域的圆形度和长宽比进行裂纹识别。实验结果表明,本方法可有效地对聚晶金刚石复合片表面裂纹缺陷进行检测。     

Recognizing people’s identity in construction sites with computer vision: A spatial and temporal attention pooling network

Several prototype vision-based approaches have been developed to capture and recognize unsafe behavior in construction automatically. Vision-based approaches have been difficult to use due to their inability to identify individuals who commit unsafe acts when captured using digital images/video. To address this problem, we applied a novel deep learning approach that utilizes a Spatial and Temporal Attention Pooling Network to remove redundant information contained in a video to enable a person’s identity to be automatically determined. The deep learning approach we have adopted focuses on: (1) extracting spatial feature maps using the spatial attention network; (2) extracting temporal information using the temporal attention networks; and (3) recognizing a person’s identity by computing the distance between features. To validate the feasibility and effectiveness of the adopted deep learning approach, we created a database of videos that contained people performing their work on construction sites, conducted an experiment, and then performed k-fold cross-validation. The results demonstrated that the approach could accurately identify a person’s identity from videos captured from construction sites. We suggest that our computer-vision approach can potentially be used by site managers to automatically recognize those individuals that engage in unsafe behavior and therefore be used to provide instantaneous feedback about their actions and possible consequences.     

一种结合区域检测和语义分割的SLAM技术

提出一种结合区域检测和语义分割的即时定位和建图（SLAM）技术，通过引入高精度图像描述子SIFT来实现前端视觉里程计（VO）过程中帧间像素匹配的精度。为了降低引入操作带来的计算复杂度，设计一个实时区域检测算法，在相邻帧间检测大致相似的ROI(Region of Interest)关键区域，使得SIFT描述子的提取和匹配只在ROI区域内完成，其余区域仍旧采用精度略低、效率更高的ORB算子。同时，为了提高后端BA(Bundle Adjustment)的精度，减少累积误差，结合语义图，在原有的基本投影误差函数上添加一个语义误差。该语义图采用实时语义分割算法完成，同时只针对ROI区域进行分割。通过与原SLAM方案对比实验，表明本文提出的方法，在提高一定精度的同时，仍能满足SLAM实时定位和建图的要求。最后，在电力作业场景下验证了该方案的效果。     

Visual aesthetic perception of handwoven cotton fabrics

We revealed the factors that govern the visual aesthetic perception of handwoven fabrics by performing two psychological experiments to determine the effects of cotton hand-spun yarn on the perceptual and cognitive responses to woven fabrics. First, a free sorting task followed by multidimensional scaling and hierarchical cluster analysis explored the effect of yarn type (hand-spun, machine-spun, and slub-yarn) on the woven fabric. Visual categorization of the fabrics depended on yarn type. Second, functional and aesthetic perceptions that produced the differences among yarn types were investigated by semantic differential scaling method that collected subjects’ ratings for the fabrics from 17 bipolar adjective pairs. Generalized linear mixed-effects modeling revealed the association between the yarn type and perception expressed by functional and aesthetic words. Visual aesthetic perception of fabrics was also affected by the consumers’ background. Our method and findings can be applied to examine the visual aesthetic of other textile products.     

基于视觉暗示的战争纪念建筑设计手法探究

当代建筑中，视觉暗示的设计手法对纪念建筑的营造有着至关重要的作用。现试图从建筑学的视角出发，以当代战争纪念建筑实际案例为研究对象，对战争纪念建筑的设计方法展开讨论。首先着重分析视觉暗示与格式塔心理学的关系，其次阐述视觉暗示给参观者心理带来的微妙变化，之后探讨视觉暗示对空间纪念效果造成的影响，最终归纳为可操作的设计策略：图形暗示、内容暗示、秩序暗示，并明确其如何相互影响。立足于格式塔心理学及心理暗示的理论基础，是对纪念建筑设计方法的补充，亦可为今后的战争纪念馆设计提供启示。     

Distortion correction and characteristics measurement of circumferential liquid film based on PLIF

For its advantages of noninvasion and high temporal–spatial resolution, planar laser-induced fluorescence (PLIF) is increasingly applied to measure the liquid film near axial flow. However, the circumferential distribution characteristics of liquid film, which are significant for heat and mass transfer process, cannot be measured briefly by the axial imaging. Because of the refractive index difference of the gas and liquid as well as the circular pipe, circumferential observation suffers distortion inevitably. In this paper, PLIF is developed to measure the circumferential film with a virtual dual-view vision sensor, besides, the distortion model is established and a new distortion correction method is proposed. The effective view of circumferential liquid film is extended to 260°. Characteristics of vertical falling film are measured under different Reynolds numbers. Compared with the classical prediction models, the experimental results show the proposed method is capable for measuring circumferential film and the distribution characteristics conveniently and reliably.     

基于小型Zynq SoC硬件加速的改进TINY YOLO实时车辆检测算法实现

针对TINY YOLO车辆检测算法计算量过大，且在小型嵌入式系统中难以达到实时检测要求的问题。利用小型Zynq SoC系统的架构优势以及TINY YOLO的网络权值中存在大量接近零的权值参数这一特点，提出硬件并行加速的改进算法，称为浓缩小型深度网络（Xerantic-TINY YOLO，X-TINY YOLO）车辆检测算法。首先对TINY YOLO中网络结构进行压缩；其次采用高效多级流水线、流水线内全并行的方式对卷积计算部分进行算法加速；最后提出与网络结构相配合的数据切割和传输方案。实验结果表明，X-TINY YOLO仅消耗50%的片内硬件资源，可在相对于GPU和CPU性价比更高更适合嵌入式场景的Zynq SoC系统上实现，且其检测速度达到24帧/s，满足车辆检测的实时性要求。     

基于多视图立体视觉的沙堆三维尺寸测量研究

利用从运动中恢复结构方法(SFM),提出了一种基于多视图立体视觉的沙堆三维重建及三维尺寸测量方法。首先根据SFM方法的求解不稳定特点,结合光束平差法对SFM求解过程进行分析及优化;其次针对SFM重建结果为稀疏点云的问题,利用基于面片的稠密重建算法重新生成稠密的三维点云,再利用泊松算法对密集点云进行三维曲面重建;最后获得模型的三维尺寸信息。对某建筑工地的沙堆进行了三维尺寸的测量实验,实验结果验证了该方法的有效及可行性,提高了重建能力及精度,同时考虑了目标实际测量误差与重建误差,能够满足实际智能测量的应用需求。