Deep learning-based methods, particularly deep convolutional neural networks (DCNNs), have demonstrated exceptional performance in extracting roads from high-resolution remote sensing images. However, current DCNNs often struggle to accurately extract small roads or roads that are substantially occluded because of the loss of position and global context information, resulting in incomplete and fragmented outputs. Graph neural networks have shown promise in modeling long-range and cross-scale features, as well as representing the overall structure of irregular objects. We propose a dual-branch fusion of convolutional and graph convolutional network (DBCGCN) for road extraction from high-resolution remote sensing images. The DBCGCN comprises two branches-a graph convolutional network (GCN) and a convolutional network-to generate complementary high-level semantic features and low-level detail features at superpixel and pixel levels, respectively. Furthermore, we employ a mathematical strategy to facilitate the exchange of features between graph nodes and image pixel points, enabling seamless collaboration between GCN and DCNN within a network. Extensive experiments conducted on widely recognized road extraction datasets validated the effectiveness of the proposed method in terms of both quality and quantity. The proposed network can improve the accuracy and reliability of road extraction in various real-world applications.

Detecting cracks from optical images plays a crucial role in road maintenance but its good realisation has many challenges. Road cracks exhibit significant diversity and complexity in terms of shape, size and texture and road images may contain various types of noise and interference, such as lighting variations, shadows and different appearances, due to varying perspectives and scales. To address these challenges, a comprehensive road crack dataset called CRCrack has been constructed, which encompasses various crack characteristics. This study proposes a road crack segmentation network called CSegNet, which combines convolutional neural networks (CNNs) and transformers. The network adopts an encoder-decoder framework, namely DeepLabV3+. In the encoder, leveraging the transformers' flexibility in modelling long-term dependencies and the CNNs' ability to capture local contextual information through local receptive fields, weight sharing and spatial subsampling, a ResNeXt-Transformer (ResNeXTR) feature extraction module is designed as the backbone network to enhance the feature extraction capability for road crack images. To reduce the computational cost in the self-attention (SA) computation of the transformer, an average pooling layer is introduced to downsample the dimensions of the encoded features. In the decoder, to focus on the key information of road cracks under diverse environmental conditions and interferences, an efficient channel attention module (ECAM) and a spatial attention module (SAM) are combined to design an efficient convolutional block attention module (ECBAM) to optimise feature representation. Through comparative experiments on the CRCrack dataset, the results demonstrate that the proposed method outperforms classic networks such as U-Net and DeepLabV3+ in terms of intersection over union (IoU), Dice coefficient and area under the receiver operating characteristic (AUROC) curve evaluation metrics. It exhibits good adaptability to ground crack images from different sources, providing a basis for estimating the degree of road damage.

Detecting cracks from images plays a crucial role in road maintenance. Road cracks exhibit significant diversity and complexity in terms of shape, size, texture, and road images may contain various noises and interferences such as lighting variations, shadows, and different appearances due to varying perspectives and scales. To address these challenges, we constructed a comprehensive dataset called the Comprehensive Road Crack Dataset (CRCrack Dataset), which encompasses various crack characteristics. In this study, we propose a road crack segmentation network called CSegNet (Crack Segmentation Network), which combines convolutional neural networks (CNNs) and Transformers. The network adopts an encoder-decoder framework, like DeepLab V3+. In the encoder, leveraging the flexibility of Transformers in modeling long-term dependencies and the ability of CNNs to capture local contextual information through local receptive fields, weight sharing, and spatial subsampling, we design a ResNeXTR (ResNeXt-Transformer) feature extraction module as the backbone network to enhance the feature extraction capability for road crack images. To reduce the computational cost in self-attention computation of transformer, we introduce an average pooling layer to downsample the dimensions of the encoded features. In the decoder, to focus on the key information of road cracks under diverse environmental conditions and interferences, we combine the Efficient Channel Attention Module (ECAM) and the Spatial Attention Module (SAM) to design an Efficient Convolutional Block Attention Module (ECBAM) attention module to further optimize feature representation. Additionally, we employ the ReLU activation function, SGD gradient descent, and a hybrid loss function of Binary Cross Entropy with Logits to accelerate convergence speed and improve segmentation accuracy. Through comparative experiments on the CRCrack dataset, the results demonstrate that our proposed method outperforms classic networks such as U-Net and DeepLab V3 + in terms of IoU, Dice, and AUROC evaluation metrics. It exhibits good adaptability to ground crack images from different sources, providing a basis for estimating the degree of road damage.

为了解烟草行业质量数据分析利用现状以及应用需求,选取卷烟、烟叶、烟用材料等12类产品对行业重点工商企业质量检测机构开展调研.结果表明:①行业质量数据分布层级清晰,数据量大,覆盖率高,能够满足大数据分析需求;②主要采用传统描述性统计方式完成数据分析,数据分析需求集中在卷烟产品的质量状况、质量趋势和预测、实验室能力水平和资源配置情况等方面;③数据分布分散且孤立,信息化系统多且系统间协同性差是存在的主要问题.在构建行业质量大数据时要重点围绕质量数据库建设、数据标准体系制定、大数据分析技术应用、数据共享与安全以及技术人才培养等方面开展工作.该研究可为充分利用质量数据资源提升卷烟产品质量提供支持.

为了全面了解卷烟表观燃烧性能质量状况,对近年来代表性规格的卷烟产品进行了燃烧过程外观形貌和落头倾向测试,基于对测试结果数据的挖掘,从产品设计和原辅材料角度对表观燃烧性能进行了分析.结果表明:①所挑选的436个市售卷烟样品总体燃烧锥落头倾向1.58％,其中中支烟落头倾向最低,细支烟最高.烟支硬度大、卷烟纸透气度60 CU、烟丝宽度1.1 mm、横罗纹卷烟纸样品落头倾向均较低.常规烟落头多发于中部靠上位置,细支烟的落头多发于偏下部位置,中支烟和短支烟均在中部靠下位置.②细支烟的包灰颜色最白、裂口率最低、缩灰率最高、燃烧匹配性较好,常规烟包灰颜色最深、裂口率最高、燃烧匹配性最差,中支烟整体较为均衡、缩灰率最低.③横罗纹、高透气度卷烟纸、烟丝宽度小、全叶丝样品的包灰性能相对较好.

为解决卷烟主要物理指标质量监控样品无法检测通风率、无法重复使用等问题,设计了一种应用于卷烟综合测试台且可重复使用的监控样品.使用弹性材料作为包裹层、多孔材料作为填充层,实现监控样品的重复使用;利用吸阻恒流管和通风率恒流管,实现卷烟吸阻和通风指标的模拟;通过质量控制图实现检测过程监控,有效识别检测中的问题.对监控样品的性能和适用性进行评价,并对主要应用场景进行效果分析,结果表明:①实现了对卷烟质量、圆周、吸阻、硬度、长度、通风率等全部物理指标的模拟检测;②监控样品的稳定性和均匀性显著优于卷烟,各项指标变异系数均明显低于卷烟样品,硬度的变异系数<2％,其余指标的变异系数均<1％,连续5年测试的吸阻极差为9 Pa,通风率极差为1.7％,其余指标平均值和标准偏差基本无变化,适用于不同型号的卷烟综合测试台;③监控样品在卷烟物理指标实验室比对、检测过程监控、标准测量方法的重复性与再现性评价、测量审核4种质量控制活动中应用效果良好,能够保证测试结果的有效性与可比性.

Natural cellulose paper is flame retarded using ammonium polyphosphate/montmorillonite (APP/MMT) nanocompound through coating method. Their morphologies are assessed by scanning electron microscopy, chemical components by energy dispersive X-ray spectroscopy and X-ray photoelectron spectroscopy, chemical structure by Fourier transform infrared, thermal stability by thermo-gravimetric analysis and fire retardancy by 45° horizontal burning test. The results show that the thermal stability and flame retardancy of coated paper samples are enhanced obviously, and all samples are self-extinguishing during the tests due to the char formation reaction between APP/MMT and cellulose. Then the treated paper, AM-18.3, as surface flame retardant treatment for low specific surface area (SSA) polypropylene (PP) specimens (PP@AM-18.3) also is investigated using cone calorimeter. Compared with two controlled samples, PP@AM-18.3 shows significantly increased time to ignition, the lowest fire growth index and highest fire performance index. This may give a new flame retarded approach for low SSA polymer materials with little amount flame retardants.

为快速、准确地识别叶丝、梗丝、膨胀叶丝、再造烟叶丝等烟丝类型,利用各类烟丝图像特征差异,以残差神经网络为基础构建了识别模型,并对模型的预训练权值、优化算法、学习率等超参数进行了研究,结果表明:①基于残差神经网络的识别方法可以有效识别4种类型烟丝,相比基于卷积神经网络的识别方法,模型具有更高的识别率、泛化能力与鲁棒性.②较优超参数对模型的训练速度及表现影响显著,通过训练得到的模型在测试集上的准确率及召回率均高于96％,且与训练集表现差异较小.该方法可为提高烟丝类型识别效率和准确性提供支持.

[目的]针对真伪卷烟包装鉴别任务对类别精度要求高、深度残差等网络不能提取出更具判别力特征的问题,本文从保证图像高分辨率表征的角度出发,提出了结合高分辨率网络(High-Resolution Network,HRNet)和注意力机制的方法,以得到更具表现力的特征,从而达到提高真伪卷烟包装鉴别准确度的目的.[方法]以具备并行子网结构的高分辨率网络为骨干网络,通过多分辨率特征融合方法获得鉴别卷烟真伪的高质量特征,并在此网络基础上嵌入了高效通道注意力(Efficient Channel Attention,ECA)模块,有效地增强了通道之间的信息交互.[结果]经过实验验证,本文提出的方法不仅可以学习到更好的特征表示,而且准确率可达到97.21％.[局限]模型着重关注了通道维度的相关性,忽略了特征空间位置信息,还有改进空间.[结论]通过将高分辨率网络和注意力机制相结合,可以有效地提高卷烟真伪鉴别的准确度,并为相关研究提供了一种新的研究思路.

为了考察环境条件对爆珠力学性能的影响,通过测试不同环境条件下爆珠的质量、压破强度、压破形变率和弹性指数,提出了爆珠样品的适宜环境条件.结果表明:①爆珠和爆珠滤棒在"温度(22±1)℃、相对湿度(60±2)％"的环境下调节48 h能够达到温湿度平衡.②爆珠的压破强度、压破形变率和弹性指数在40％～60％相对湿度范围和10～40℃温度范围内略有差异,但差异的程度与样品差异有较大关系.极限温度条件(低于10℃,高于40℃)对爆珠的力学性能有较大的影响,低温冷冻后爆珠的力学性能不可恢复.③爆珠的力学性能在爆珠生产、爆珠滤棒加工和爆珠卷烟加工各环节的温湿度条件下均没有较大差异.

The cigarette detection data contains a large amount of true sample data and a small amount of false sample data.The false sample data is regarded as abnormal data, and anomaly detection is performed to realize the identification of real and fake cigarettes.Binary particle swarm optimization algorithm is used to improve the isolation forest construction process, and isolation trees with high precision and large differences are selected, which improves the accuracy and efficiency of the algorithm.The distance between the obtained anomaly score and the clustering center of the k-means algorithm is used as the threshold for anomaly judgment.The experimental results show that the accuracy of the BPSO-iForest algorithm is improved compared with the standard iForest algorithm.The experimental results of multiple brand samples also show that the method in this paper can accurately use the detection data for authenticity identification.

为了考察样品放置方式对卷烟温湿度调节效果和主要物理指标的影响,研究了散装和盒装剪底两种常用样品放置方式对卷烟质量平衡状态和卷烟单支质量、吸阻、总通风率等指标的影响.结果表明:①不同样品放置方式下,卷烟含水率和达到质量平衡的时间有较大差异,散装卷烟经48 h调节后达到平衡状态,盒装剪底调节的卷烟质量随调节时间增加持续增加.②样品放置方式对卷烟不同物理指标的影响有一定差异,对卷烟硬度的影响最大,会造成产品硬度判定结果的变化;对卷烟单支质量有较大的影响;对卷烟吸阻、总通风率、圆周、长度的影响较小.相同调节时间内散装调节的卷烟的质量、吸阻、总通风率、圆周、长度均大于盒装剪底调节的卷烟,硬度小于盒装剪底调节的卷烟.③烟草及烟草制品调节和测试的大气环境标准中应进一步明确卷烟样品温湿度调节时的放置方式,以避免其差异对测试结果的影响,建议调节卷烟样品时应尽量采用散装方式放置.

针对实验室检测能力建设,遵循普遍适用的设计理念,浅述实验室检测能力设计的必要性、原则和相关步骤,阐述了设计是保证检测结果质量的前提,并从方法选择、测量原理的理解、试验、资源和供给、检测流程设计、控制策略、实验等方面,介绍了如何通过规划、方案设计、详细设计、改进设计等阶段进行检测能力的设计.

为满足卷烟纸产品属性鉴别检验的要求,研究了卷烟纸与其他外观形态相似的薄页纸张的阴燃性能和助燃剂钾元素含量,通过对比实验分析了各类纸张的差异,并验证了方法的可行性.结果表明,卷烟纸和其他外观形态相似的普通薄页纸张的燃烧状态、阴燃长度和助燃剂钾元素含量均有明显差异,能够通过阴燃性能和钾元素含量区分卷烟纸和其他外观形态相似的普通薄页纸张.

[目的]利用改进的尺度不变特征变换(Scale-Invariant Feature Transform,SIFT)算法提取的匹配特征对卷烟商标纸图像进行细粒度配准,达到提升配准精度和区分真伪卷烟商标纸图像的目的.[方法]通过对图像分块处理、剔除不稳健特征点、单应性矩阵粗配准后根据匹配点距离进行约束筛选匹配对,并提出根据细粒度配准后的匹配点距离均值进行评价,最终实现并改进了基于特征点的卷烟商标纸细粒度图像配准方法.[结果]基于本文改进的特征点检测方法可以提取到更均衡的特征点,提高推定匹配率,提出的配准结果评估标准能有效评估配准质量,粗配准筛选匹配点可以提高图像细粒度配准的精度,并可以对卷烟商标纸图像进行区分.[局限]目前的改进集中在匹配对的筛选,在细粒度配准方法研究上仍有改进的空间.[结论]基于改进的SIFT算法提取的特征点,提出了先粗配准后细配准的图像细粒度配准策略,经实验证明此策略可以提升图像配准精度,并可以达到区分卷烟商标纸图像的目的.

为解决人工鉴别真伪卷烟效率低、主观性强等问题,基于计算机视觉和机器学习建立了一种真伪卷烟包装鉴别模型.利用计算机视觉对卷烟包装进行图像处理和特征向量提取,分别以相似性度量模型、机器学习模型对特征向量进行分类并判定卷烟真伪.相似性度量模型采用曼哈顿距离模型进行分类,并对高斯双边滤波函数进行了参数优化;机器学习模型则以图像分块为基础,确定最优分块数量和面积.以"中华(软)""玉溪(软)""钻石(荷花)"3个卷烟品牌共603个真伪样品为对象,分别采用两种模型进行判定,结果表明:相似性度量模型在"玉溪(软)"样品测试集的准确率为96.17％;机器学习模型在"中华(软)""玉溪(软)""钻石(荷花)"3个样品测试集的准确率分别为98.99％、96.61％和100％.机器学习模型与相似性度量模型相比较,具有较好的迁移能力和鲁棒性,适用于卷烟真伪鉴别样品量大、品类多、图像复杂等情况.该方法可为提高真伪卷烟鉴别效率和准确率提供技术支持.

为了探索新型阻燃剂(LD及APP)对低引燃倾向(LIP)卷烟纸阻燃性能的影响.分别将阻燃剂LD、阻燃剂APP以及二者复配制备成相应的阻燃剂涂料,利用涂布的方法在卷烟纸上涂布阻燃带生产LIP卷烟纸,并对用该卷烟纸卷制的LIP卷烟样进行全长燃烧比例及阻燃带扩散率的测定.结果 表明,当单独使用阻燃剂LD或APP时,两者都具有阻燃性,LD的阻燃性能优于APP;但单独使用LD进行涂布时,阻燃带存在边界扩散渗透现象;且随着LD含量的增大,阻燃带上的LD容易出现脱落(掉粉)现象.LD与APP进行复配可解决上述问题.当LD与APP复配阻燃剂(LD与APP质量比7∶3)含量达到16％左右时,卷烟的全长燃烧比例为20％,卷烟纸阻燃带扩散率为1.000cm/s,可满足LIP卷烟纸的性能要求.

In order to assess the cutting quality of stem quickly, an analysis method based on visual morphological feature detection was proposed. Morphological analysis method was used to compute the shred content in cut stems (the proportion of shred in cut stems);the length of cut stem was computed by skeleton extraction method and its uniformity was computed by probability density function, the width of cut stem was computed by Hough transform principle and its uniformity by numerical analysis. The proposed method was verified by accuracy, precision and efficiency experiments. The results showed that the measurement accuracy of shred content in cut stems reached 98.73%, the measurement errors of both cut stem length and width were 0.02, which indicated that the measurement was accurate. The variation coefficients of measured shred content in cut stems, length uniformity and width uniformity were 0.6%, 1.0% and 1.5%, respectively;the measurement had good precision. The average time needed for measuring cut stems of 0.01 kg was 11 s with the image processing time less than 2.5 s, which satisfied the requirements for on-line measurement of cut stems of large quantities.

In order to promote the efficiency and accuracy for detecting capsule in filter rod, a detector was designed based on microwave resonance cavity perturbation technique. The detector was mainly composed of a filter rod feeding device, a filter rod conveying device, a microwave detection unit and a sorting device. The capsule missing, position and defects of capsule, in filter rod could be detected quickly via an established wave crest shape and position recognition algorithm. The results showed that the designed detector and the established algorithm featured higher accuracy for detecting the position of capsule and good recognition effect for defects or absence of capsule. The maximum standard deviation was 0.111 mm in the capsule position repeatability test, and the accuracy of capsule defect detection reached 100% at a rate of 120 rods per minute. The detector provides a rapid and accurate quality detection method for capsuled-filter rods.