多任务学习孪生网络的遥感影像多类变化检测

精确掌握土地覆盖/利用的变化及变化类型对国土空间规划、生态环境监测、灾害评估等有着重要意义,然而现有大部分变化检测研究主要关注二值变化检测。为此,该文首先提出了一种多任务学习深度孪生网络用于遥感影像的多类变化检测。首先提出面向对象的无监督变化检测方法,选择出新、旧时相影像中最有可能发生变化和最不可能发生变化的区域,并作为多任务学习深度孪生网络的样本;其次,采用多任务学习深度孪生网络模型同时对新、旧时相的土地利用图以及新、旧时相的二值变化图这3个任务模型进行学习和预测;最后,基于模型预测的新、旧时相土地利用图及新、旧时相的二值变化图获取最终的多类变化检测结果。采用第三次全国国土调查的影像数据和相应的土地利用图斑数据对多任务学习深度孪生网络模型进行了测试,结果表明所提出的方法适用于这种在没有变化、未变化样本而有历史专题图的变化检测场景中。

结合重压缩检测的JPEG图像多类隐写分析

提出一种新的结合重压缩检测的JPEG图像多类隐写分析方法,实现一次压缩和重压缩图像中多种隐写算法的识别.首先基于DCT系数首位数分布规律,提出一种重压缩检测方法,然后从系数直方图、块内相关性、块问相关性和空域块效应中提取盲检测特征用于隐写分析,最后用支持向量机构造JPEG隐写算法多类检测器.实验结果表明,本文方法的重压缩检测性能明显优于已有方法,且对嵌入改变量的鲁棒性较强,隐写分析特征不仅维数较低而且具有更好的检测性能,构造的多类隐写分析器能较好地识别JPEG隐写算法.

RSVP与SSVEP混合脑电信号刺激与多类事件检测

提出一种新的基于快速序列视觉呈现(Rapid Serial Visual Presentation,RSVP)与稳态视觉诱发电位(Steady-State Visually Evoked Potential,SSVEP)组合范式的脑电信号(Electroencephalogram,EEG)刺激与多类事件检测方法。对诱发的原始脑电信号通过电位重参考、基线去除、空间滤波等预处理操作去除数据的伪迹和噪声,通过自举聚合决策树(Bagging Tree,BT)和支持向量机(Supported Vector Machine,SVM)等机器学习算法,对14名受试者双重刺激诱发的脑电信号进行目标与频率相结合的多类事件检测,通过实验验证了该组合范式诱发的脑电信号具有良好的多类可分性,为开发基于RSVP和SSVEP两种范式的混合型脑-机接口应用提供了一种新的有效途径。同时,实验结果还表明,基于机器学习的BT和SVM模型对RSVP和SSVEP组合范式诱发的EEG信号进行多类识别的性能明显优于传统的典型关联分析(Canonical Correlation Analysis,CCA)算法的性能。

自适应结构化背景和形状特征子空间高光谱图像多类目标检测

针对高光谱图像中目标形状特征已知,背景和目标光谱特征未知时的多类小目标检测问题,给出一种检测算法.通过高光谱图像数据样本二次型的高阶矩控制点扩散函数,获取自适应结构化背景;然后,利用目标形状先验信息构造形状特征子空间,在高维光谱特征空间实现形状特征子空间匹配检测.理论分析和实验结果表明该检测器可同时有效检测具有不同形状特征的多类目标.

基于多类攻击的Modbus Tcp网络入侵检测方法

为了提高Modbus/Tcp网络安全水平,减少异常攻击,文中提出一种基于多类攻击的Modbus/Tcp网络入侵检测方法。该方法首先通过核主成分分析算法(KPCA)降低冗余数据,然后使用决策树算法(C4.5)创建多类分类器识别异常攻击行为以及小样本攻击,最后对未知攻击数据使用k-邻近(k-NN)值提升精度。实验对4类攻击进行精度与误报率的对比,结果表明,该方法不仅可以大大缩短样本训练的时间,而且在后期的识别攻击行为的实验中,相对于当前主流的检测具有更好的精度、误报率和更短的时间效率。

基于改进YOLOv5s和多模态图像的树上毛桃检测

毛桃等果实的准确检测是实现机械化、智能化农艺管理的必要前提。然而,由于光照不均和严重遮挡,在果园中实现毛桃,尤其是套袋毛桃的检测一直面临着挑战。本研究基于改进YOLOv5s和多模态视觉数据提出了面向机械化采摘的毛桃多分类准确检测。具体地,构建了一个多类标签的裸桃和套袋毛桃的RGB-D数据集,包括4127组由消费级RGB-D相机获取的像素对齐的彩色、深度和红外图像。随后,通过引入方向感知和位置敏感的注意力机制,提出了改进的轻量级YOLOv5s(小深度)模型,该模型可以沿一个空间方向捕捉长距离依赖,并沿另一个空间方向保留准确的位置信息,提高毛桃检测精度。同时,通过将卷积操作分解为深度方向的卷积与宽度、高度方向的卷积,使用深度可分离卷积在保持模型检测准确性的同时减少模型的计算量、训练和推理时间。实验结果表明,使用多模态视觉数据的改进YOLOv5s模型在复杂光照和严重遮挡环境下,对裸桃和套袋毛桃的平均精度(Mean Average Precision,mAP)分别为98.6%和88.9%,比仅使用RGB图像提高了5.3%和16.5%,比YOLOv5s提高了2.8%和6.2%。在套袋毛桃检测方面,改进YOLOv5s的mAP比YOLOX-Nano、PP-YOLO-Tiny和EfficientDet-D0分别提升了16.3%、8.1%和4.5%。此外,多模态图像、改进YOLOv5s对提升自然果园中的裸桃和套袋毛桃的准确检测均有贡献,所提出的改进YOLOv5s模型在检测公开数据集中的富士苹果和猕猴桃时,也获得了优于传统方法的结果,验证了所提出的模型具有良好的泛化能力。最后,在主流移动式硬件平台上,改进后的YOLOv5s模型使用五通道多模态图像时检测速度可达每秒19幅,能够实现毛桃的实时检测。上述结果证明了改进的YOLOv5s网络和含多类标签的多模态视觉数据在实现果实自动采摘系统视觉智能方面的应用潜力。

多核多类关联向量机的高分辨率影像目标检测

从高分辨率遥感图像数据中准确检测多类目标的任务对于检测速度和模型训练时间提出了较高的要求。文章提出了一种MKL_mRVM方法:该方法采用基于快速边缘似然最大算法直接计算mRVM分类器的决策函数,避免了传统RVM重复计算目标函数Hessian矩阵的过程,并且因为不需要构造一系列两类分类器,缩短了多类模型的训练时间;同时,将多个基础核引入多类模型,训练过程中采用交叉验证方法确定基础核权重,在随机分出的确认集上检验分类器的精度,选取使得分类模型精度最高的值作为权重的优化结果。实验结果表明,该方法能够在保持解的稀疏性的前提下,有效地缩短模型训练时间。

色质联用技术的进步与农药多残留分析方法的发展(之一)

气相色谱-质谱联用技术是农药多类多残留检测分析方法发展的起点,液相色谱-质谱联用技术的发展给极性农药、热不稳定性农药及其代谢物的检测创造了条件,高选择性、高灵敏度、高通量的色谱-串联质谱技术创建了农药的多残留检测方法。本文回顾了色质联用技术的进步及其在农药多残留分析方法发展过程中的重要作用。

Distribution of Pill Millipedes (Arthrosphaera) and Associated Soil Fauna in the Western Ghats and West Coast of India

Seven sampling sites in each of three biomes (Western Ghats, foothills of Western Ghats and west coast) of south- western India were investigated to study the distribution, abundance and ecology of pill millipedes (Arthrosphaera) and associated fauna in relation to edaphic features. Abundance and biomass of Arthrosphaera and other millipedes were the highest in Western Ghats, while earthworms were in foothills. Arthrosphaera magna and Arthrosphaera spp. were common in Western Ghats and foothills respectively, while no Arthrosphaera were found in the west coast. None of the sampling sites consisted of more than one species of Arthrosphaera. Biomass of Arthrosphaera, other millipedes and earthworms significantly differed in Western Ghats (P = 9.48 × 10-7) and foothills (P = 1.35 × 10-8), as did the biomass of species of Arthrosphaera (P = 2.76 × 10-7) between Western Ghats and foothills. Correlation analysis revealed that biomass of Arthrosphaera was significantly (P = 0.01, r = 0.45) correlated with soil organic carbon rather than other edaphic fea- tures (pH, phosphate, calcium and magnesium). Distribution pattern, abundance, biomass and ecology of Arthrosphaera of Western Ghats in relation to soil qualities were compared with millipedes of other regions of the world.

Length-Based Vehicle Classification in Multi-lane Traffic Flow

For the realtime classification of moving vehicles in the multi-lane traffic video sequences, a length-based method is proposed. To extract the moving regions of interest, the difference image between the updated background and current frame is obtained by using background subtraction, and then an edge-based shadow removal algorithm is implemented. Moreover, a tbresholding segmentation method for the region detection of moving vehicle based on lo- cation search is developed. At the estimation stage, a registration line is set up in the detection area, then the vehicle length is estimated with the horizontal projection technique as soon as the vehicle leaves the registration line. Lastly, the vehicle is classified according to its length and the classification threshold. The proposed method is different from traditional methods that require complex camera calibrations. It calculates the pixel-based vehicle length by using uncalibrated traffic video sequences at lower computational cost. Furthermore, only one registration line is set up, which has high flexibility. Experimental results of three traffic video sequences show that the classification accuracies for the large and small vehicles are 97.1% and 96.7% respectively, which demonstrates the effectiveness of the proposed method.

The detection and significance of polycyclic aromatic hydrocarbons in human pancreatic cancer

Objective:The aim of this study was to investigate the effect of polycyclic aromatic hydrocarbons (PAHs) on the occurrence of human pancreatic cancer.Methods:PAHs in human pancreatic cancer,adjacent pancreatic cancer tissues and tissues without pancreatic cancer were extracted by ultrasonic extraction (UE).And then the extracts were cleaned up by solid phase extraction and analyzed by high performance liquid chromatography (HPLC) with fluorescence spectroscopy.Results:Four kinds of PAHs were detected,which were chrysene,2-methylanthracene,pyrene and benzo (a) pyrene.The contents of the four PAHs were not statistically significant between pancreatic cancer and adjacent tissues (P > 0.05).The contents of 2-methylanthracene,pyrene and benzo (a) pyrene in pancreatic cancer and adjacent tissues were higher than tissues without pancreatic cancer,the differences were statistically significant (P < 0.05).The contents of chrysene in the three kinds of pancreatic tissues were not statistically significant (P > 0.05).Conclusion:PAHs were found in human pancreatic tissues.Human pancreatic tissues have extremely strong ability of bio-concentrating PAHs.PAHs might play an important role in the occurrence of human pancreatic cancer.