基于LRC与多样本扩充的指静脉识别方法北大核心CSCD

针对目前手指静脉识别由于训练样本不足引起图像识别率低的问题,提出基于线性回归分类(linear regression classification,LRC)与多样本扩充的指静脉识别方法。首先,利用矩阵变换生成原始图像的镜像,训练原始图像与镜像,增加指静脉图像中包含的有用信息;然后,基于LRC对测试和训练样本进行分类;最后,通过计算偏差得到最终分类结果,求出识别率。此外,设计了一种指静脉采集装置收集得到自建指静脉数据库。实验结果表明:所提算法在自建指静脉数据库、山东大学指静脉数据库、马来西亚理工大学指静脉数据库上的识别率分别达到98.93%、98.89%、99.67%,最低等误率为2.3888%。实验结果与其他传统和流行算法相比具有明显优势,拥有良好的实际应用价值。     

多尺度混合注意力胶囊网络的海洋鱼类识别

针对胶囊网络(capsule network,CapsNet)特 征提取结构单一和数据处理中参数量过大的问题,提出 多尺度混合注意力胶囊网络 模型。首先,在网络初始端添加不同尺度的卷积核来多角度提取 特征,并引 入混合注意力机制,通过聚焦更具分辨性的特征区域来降低复杂背景干扰。其次,采用局部 剪枝算法优 化动态路由,减少参数量,缩短模型训练时间。最后,在海洋鱼类数据集F4K(Fish4Knowled ge)上验证, 结果表明,与传统残差网络(residual network50,ResNet-50)、双线性网络(bilinear convo l utional neural network,B-CNN)、分层精简双线性注意力网络(spatial transformation netw ork and hierarchical compact bilinear pooling,STN-H-CBP)以及CapsNet模型相比,该算法 识别精度为98.65%,比ResNet-50模型提升 了5.92%;训练时间为2.2 h,相比于CapsNet 缩短了近40 min,验证了该算法的可行性。     

Saliency guided self-attention network for pedestrian attribute recognition in surveillance scenarios

Pedestrian attribute recognition is often considered as a multi-label image classification task. In order to make full use of attribute-related location information, a saliency guided sel-attention network ( SGSA-Net) was proposed to weakly supervise attribute localization, without annotations of attribute-related regions. Saliency priors were integrated into the spatial attention module ( SAM ). Meanwhile,channel-wise attention and spatial attention were introduced into the network. Moreover, a weighted binary cross-entropy loss ( WCEL) function was employed to handle the imbalance of training data. Extensive experiments on richly annotated pedestrian ( RAP) and pedestrian attribute ( PETA) datasets demonstrated that SGSA-Net outperformed other state-of-the-art methods.     

Recognition of guests bearing donor and acceptor hydrogen bonding groups by heteroditopic calix[4]arene receptors

The synthesis of new hosts specifically designed for the recognition of neutral guests bearing donor-acceptor hydrogen bonding groups is described. These hosts are characterized by the presence of two distinct binding region in close proximity: the rigid π-donor cavity and the H-bond donor N-methylene-N′-phenylureido group inserted onto the upper rim of the calix[4]arene skeleton. The binding abilities of these receptors were investigated toward a series of neutral ditopic organic molecules in CDCl₃ solution by ¹H NMR spectroscopy. The results obtained show that rigidity of the calix[4]arene apolar cavity is the control element in determining efficiency. In fact, compared with the more rigid 2, host 10, where the rigidity of the cone structure is maintained by hydrogen bonding of the OH of the lower rim, a decrease of efficiency of almost one order of magnitude was observed. The cooperative effect of the two binding region of host 2 was verified with different classes of ditopic guests. Good efficiency in the recognition of urea derivatives and dimethylsulfoxide was achieved.     

三维主体分子的合成及阴离子识别研究进展

生物功能体系的模拟是生物有机化学和超分子化学发展的前沿领域之一。本文简要论述新型大二环．大三环、笼型环番以及其它三维主体分子的合成及对阴离子选择性识别作用的研究新进展。     

Molecular Recognition and Crystal Energy Landscapes: An X‐ray and Computational Study of Caffeine and Other Methylxanthines

We introduce a new approach to crystal‐packing analysis, based on the study of mutual recognition modes of entire molecules or of molecular moieties, rather than a search for selected atom–atom contacts, and on the study of crystal energy landscapes over many computer‐generated polymorphs, rather than a quest for the one most stable crystal structure. The computational tools for this task are a polymorph generator and the PIXEL density sums method for the calculation of intermolecular energies. From this perspective, the molecular recognition, crystal packing, and solid‐state phase behavior of caffeine and several methylxanthines (purine‐2,6‐diones) have been analyzed. Many possible crystal structures for anhydrous caffeine have been generated by computer simulation, and the most stable among them is a thermodynamic, ordered equivalent of the disordered phase, revealed by powder X‐ray crystallography. Molecular recognition energies between two caffeine molecules or between caffeine and water have been calculated, and the results reveal the largely predominant mode to be the stacking of parallel caffeine molecules, an intermediately favorable caffeine–water interaction, and many other equivalent energy minima for lateral interactions of much less stabilization power. This last indetermination helps to explain why caffeine does not crystallize easily into an ordered anhydrous structure. In contrast, the mono‐ and dimethylxanthines (theophylline, theobromine, and the 1,7‐isomer, for which we present a single‐crystal X‐ray study and a lattice energy landscape) do crystallize in anhydrous form thanks to the formation of lateral hydrogen bonds.     

计算机模式识别法对甲状腺机能亢进症微量元素谱的研究

据Mahlanobis距离判别法，利用计算机多元统计判别程序，对14例甲亢思者和21例健康人的血清中锌、铁、铜、锰、镁和锶等元素浓度差异进行分类判别研究。选择锌、铁、钙、锶作判别特征参量时，患者和健康人的分类准确率为100％，服碘前与服碘后的甲亢患者分类准确率为92.7％，分类研究指出，甲亢患者血清中的铁和锶浓度高于健康人，经服碘治疗后，患者血清中上述元素浓度降低与健康人相近。     

Visual recognition of melamine in milk via selective metallo-hydrogel formation

A novel protocol allowing convenient and highly selective visual recognition of melamine in raw milk via selective metallo-hydrogel formation at a concentration as low as 10 ppm without any tedious pretreatment has been developed.     

New chemosensor for larger guests based on modified cyclodextrin bearing seven hydrophobic chains each with a hydrophilic end group

A new chemosensor for larger guests was prepared. The new chemosensor bears hydrophobic units at the primary hydroxy side and a dansyl unit at the secondary hydroxy side of β-cyclodextrin. Due to the hydrophobic units, the new chemosensor is sensitive to large or slender guests such as SDS and insensitive to 1-adamantanol, which is a good guest for the natural β-CD.     

Crystal Structures of 1,5,9,18,22,26-Hexaaza[11.11]-p-cyclophane Adducts; Two-dimensional Supramolecular Networks

Two newly identified supramolecular structures arise from self-assembly of the macrocyclic 1,5,9,18,22,26- hexaaza[11.11]-p-cyclophane salts with o-nitrophenol (C₂₈H₅₀N₆)⁴⁺·4(C₆H₄NO₂O)⁻ (1) and with HCl (C₂₈H₅₂N₆)⁶⁺·6Cl^-·4H₂O (2). In both cases two-dimensional supramolecular sheets are formed.