基于改进随机漫步者非凸秩逼近最小化的复杂背景下红外点目标检测方法研究

红外点目标检测是红外制导系统的关键技术之一,是军事应用领域的研究热点。一方面,点目标在大气传输和散射过程中由于观测距离长,常常淹没在背景杂波和大噪声中,信噪比低。另一方面,图像中的目标以模糊点的形式出现,使得目标没有明显的特征和纹理信息。由于不同红外图像中的点目标具有不同的外观、形状和姿态,加之噪声杂波的干扰遮挡,经过单帧检测后,除了真实目标外,图像中可能有虚假目标和一些强噪声。因此,由于这几个因素,红外点目标检测变得非常困难。为了解决这一问题,作者研究了红外点目标检测的相关方法,提出了非凸秩逼近最小化方法（NRAM）与改进的随机漫步者方法（MRW）相结合的方法（NRAM-MRW）,在复杂的红外背景下,针对红外点目标的检测中有着较好的检测效果。     

基于sMLD特征的远红外航空图像配准算法

航空场景下的远红外图像迫切需要准确、鲁棒、快速的特征描述与自动配准方法。由于已有的多重线型描述符MLD特征存在“特征孤岛”和“尺度变换受限”的问题,因此针对远红外航空图像的配准问题,文章提出了一种聚合特征点以及线型描述符分段统计的特征描述方式,即sMLD特征。结合sMLD特征相互连接构成网状拓扑结构的特性,还提出了一种由粗到细的分支加速匹配算法RF-BA。其中,RF-BA粗匹配充分利用拓扑图的结构,通过局部寻优算法提高匹配的效率。RF-BA精匹配利用最小外接凸四边形原则和GMS校验原则,提升配准精度。实验结果表明,与已有的几种代表性配准方法相比,所提方法在配准精度和时间开销方面均具有更好的性能。     

基于直线运动机构的大变倍比红外变焦成像系统

针对同时兼顾大范围搜索和精确识别目标的迫切需求,研制了一种大变倍比红外变焦成像系统,设计两片独立运动的变倍镜及一片补偿镜,通过两个变倍镜级联的方式获得大变倍比。结合系统运动镜片多及变焦曲线复杂的特点,采用直线运动机构实现镜片变焦运动,使用集成编码器及螺纹丝杆的直线电机作为驱动。通过有限元仿真开展了系统力学分析,所设计镜片最大位移为3.04×10^-3 mm。成像系统适用于中波红外制冷式640×512焦平面阵列探测器,变倍比达到55倍。实验室成像及外场实景成像的结果表明,系统在焦距由6 mm至330 mm连续变化的过程中成像清晰、像质良好,验证了系统的连续变焦成像性能,该设计合理可靠。研究成果在搜索、跟踪、侦察、监视等方面有广阔的应用前景。     

基于自适应对比度增强的红外小目标检测网络

红外探测系统需要尽早发现目标以便及时拦截,但是红外图像上的小目标检测是一个挑战十足的任务。为了提高检测准确率,提出一种基于自适应对比度增强的红外小目标检测方法。为了利用自注意力机制和卷积各自的优势,设计了一个高效的特征提取网络和一个面向小目标的检测头。同时为了解决实际应用中出现的弱目标,在检测子网络前添加了一个图像预处理子网络,该模块可以自适应地调节图像对比度。在红外空中小目标数据集上的实验表明,提出的方法能达到93.76%的检测精度,与经典的检测方法相比,能够更好地平衡检测精度和召回率,证明了方法的巨大应用潜力。     

基于梯度能带结构的高速非制冷中波红外HgCdTe探测器

报道了基于梯度能带结构的高速室温中波红外HgCdTe器件,器件设计为n-on-p同质结结构,在300 K的零偏压条件下达到了1.33 ns（750 MHz）的总的响应时间,相对于非制冷的碲镉汞器件和工作于高偏压下的碲镉汞APD器件响应速度有所提高。基于一维模型的分析表明,吸收层中的组分梯度可以形成内置电场并改变了载流子的输运特性,该模型由不同组分梯度的HgCdTe器件的实验对比验证。因此,此项工作优化了高速HgCdTe中波红外探测器的设计,并为设计超快中波红外光电探测器提供了一种可行的思路。     

Visible–infrared person re-identification based on key-point feature extraction and optimization

Feature extraction for visible–infrared person re-identification (VI-ReID) is challenging because of the cross-modality discrepancy in the images taken by different spectral cameras. Most of the existing VI-ReID methods often ignore the potential relationship between features. In this paper, we intend to transform low-order person features into high-order graph features, and make full use of the hidden information between person features. Therefore, we propose a multi-hop attention graph convolution network (MAGC) to extract robust person joint feature information using residual attention mechanism while reducing the impact of environmental noise. The transfer of higher order graph features within MAGC enables the network to learn the hidden relationship between features. We also introduce the self-attention semantic perception layer (SSPL) which can adaptively select more discriminant features to further promote the transmission of useful information. The experiments on VI-ReID datasets demonstrate its effectiveness.     

Comparative FT infrared study of structural changes occuring on dealumination of zeolite Na-Y with ethylenediaminetetraacetic or hydrochloric acid

IR lattice vibration spectra were used to monitor the changes in the structure of zeolite Na-Y upon its dealumination with ethylenediaminetetraacetic acid (EDTA) or with HCl. The terminal Si-O(H,Na) bonds thus formed (which are detectable by characteristic absorbance at v 900–950 cm^–1) are energetically less uniform when EDTA has been used as the dealuminating agent. This inhomogeneity is connected with the local deformations of zeolite structure which result in lowering the symmetry of silicon-oxygen tetrahedra. As a consequence, two novel bands appear at v 1090 cm^–1 and 1200 cm^–1 in addition to the fundamental absorption bands, v_as (TO₄) (T = Si, Al), at 1030 cm^–1 and 1145 cm^–1. A mechanism of dealumination is proposed, which takes into account the topochemical peculiarities involved in the chelation of the framework aluminum ions with EDTA.Translated fromIzyestiya Akademii Nauk. Seriya Khimicheskay, No. 2, pp. 284–287, February, 1993.     

同时测定水溶液中葡萄糖、果糖和蔗糖的近红外光谱法

通过对27个葡萄糖、果糖和蔗糖水溶液的混合体系进行近红外光谱分析，建立了葡萄糖、果糖和蔗糖含量测定的偏最小二乘法(PLS)模型；葡萄糖、果糖和蔗糖的线性范围分别为0—300g/L、0—200g/L、0—300g/L，模型校正集的标准误差(SEC)分别为1．4、1．8、1．4g/L；用该模型对6个样品进行分析，葡萄糖、果糖和蔗糖含量测定的相对标准偏差(RSD)分别为1．2％、2．6％和1．8％。     

Use of NIRS technology with a remote reflectance fibre-optic probe for predicting major components in bee pollen

In the present work, we study the use of near infra-red spectroscopy (NIRS) technology together with a remote reflectance fibre-optic probe for determination of the major components in bee pollen. The method allows immediate control of the bee pollen without prior sample treatment or destruction through direct application of the fibre-optic probe to the sample.The regression method employed was modified partial least squares (MPLS). The calibration results obtained using 45 samples of bee pollen allowed the measurement of protein, moisture, ash, reducing sugars, and pH with multiple correlation coefficients (RSQ) and prediction corrected standard errors (SEPC) of 0.91, 0.56% for protein, of 0.78 and 0.49% for moisture; 0.92 and 0.049% for ash; 0.81 and 1.32 g of glucose/100 g of bee pollen; 0.84 and 0.15 for pH, respectively.The prediction capacity of the pattern was checked by applying it to samples of unknown pollen in external validation.     

Composition and major odorous compounds of the essential oil of Bifora radians,an aldehyde-producing weed

The composition of the essential oil of Bifora radians, an aldehyde-producing weed, has been investigated by capillary gas chromatography, coupled gas chromatography – mass spectrometry, on-line catalytic hydrogenation and coupled gas chromatography – infrared spectrometry. The nineteen compounds identified included eighteen aldehydes: seven alkanals (C6, C9, C10, C11, C12, C13, and C14), ten alkenals, including five (E)-2-alkenals (C12, C13, C14, C15, and C16), and one (E,E)-2,4-alkadienal (C13). Typical Bifora odors were attributed to three major (E)-2-alkenals, C12, C13, and C14.