结合坐标注意力与自适应残差连接的logo检测方法

Logo检测在品牌识别和知识产权保护等领域有着广泛的应用. 针对logo检测中存在小尺度Logo检测性能差和logo定位不准的问题, 本文提出一种基于YOLOv4网络的logo检测方法, 将YOLOv4网络PANet模块中的5个连续卷积层用设计的自适应残差块替换, 增强浅层和深层的特征利用, 有侧重地进行特征融合, 同时优化网络训练; 并在自适应残差块之后使用坐标注意力机制, 通过精确的位置信息对通道关系和长期依赖性进行编码, 从融合的特征中过滤和增强对于检测更有用的特征; 最后采用K-means++聚类算法得到更适合logo数据集的先验框, 并分配给不同的特征尺度. 实验结果表明, 本文提出的方法在FlickrLogos-32和FlickrSportLogos-10数据集上的平均精度达到了88.09%和84.72%, 较原算法分别提高了0.91%和1.40%, 在定位精度和小尺度logo检测上的性能都显著提升.     

Estimating parameters of a two-layer stratified ocean from polarity conversion of internal solitary waves observed in satellite SAR images

This paper presents a method for estimating parameters of a two-layer stratified ocean using satellite SAR images. According to weak nonlinearity shallow water theory, internal solitary waves (ISWs) in stratified oceans may be either depression or elevation waves, depending on the sign of the quadratic nonlinearity coefficient in the KdV equation. It has been confirmed that ISWs can convert their polarity when passing through a turning point, where the quadratic nonlinearity coefficient changes sign. For a two-layer stratified ocean, the turning point is located where the upper and lower layer depths are equal. The authors suggest that depression, elevation and broadening ISWs can be discerned according to their different signatures in SAR images. It is also found that a SAR image can record a continuous evolution process from depression to elevation ISWs in its spatial domain, under conditions of a spatially inhomogeneous ocean environment. Therefore, the upper and lower layer depths can be calculated by determining the polarity conversion of ISWs observed in satellite SAR images. Furthermore, the density difference between the upper and lower layers can also be estimated, when the wave speed is known. We extract ocean stratification parameters, including upper layer depth and density difference, from polarity conversion of ISWs observed in a RADARSAT-1 SAR image taken over the northeastern South China Sea. Comparing the estimated results with field measurements, we find that this method can estimate the upper layer depth with considerable success. In estimating the density difference between the upper and lower layers, it also gives a quite reasonable result.     

Hatching eggs classification based on deep learning

In order to realize the fertility detection and classification of hatching eggs, a method based on deep learning is proposed in this paper. The 5-days hatching eggs are divided into fertile eggs, dead eggs and infertile eggs. Firstly, we combine the transfer learning strategy with convolutional neural network (CNN). Then, we use a network of two branches. In the first branch, the dataset is pre-trained with the model trained by AlexNet network on large-scale ImageNet dataset. In the second branch, the dataset is directly trained on a multi-layer network which contains six convolutional layers and four pooling layers. The features of these two branches are combined as input to the following fully connected layer. Finally, a new model is trained on a small-scale dataset by this network and the final accuracy of our method is 99.5%. The experimental results show that the proposed method successfully solves the multi-classification problem in small-scale dataset of hatching eggs and obtains high accuracy. Also, our model has better generalization ability and can be adapted to eggs of diversity.     

基于门限联合判别的同质网络入侵检测仿真

动态同质网络在路由转发控制中节点分布具有动态性,容易受到病毒入侵,提出基于上下门限联合判别的动态同质网络入侵检测算法。以时间均值和谱密度为网络入侵检测的统计特征量,采用宽平稳随机序列分析方法构建网络入侵的统计信号分析模型,对入侵信号采用相关性检测和同态匹配滤波方法进行降噪和盲源分离处理,结合极速学习方法进行动态同质网络的入侵特征量提取,采用上下门限联合判别方法实现动态同质网络的入侵检测。仿真结果表明,采用该方法进行动态同质网络的入侵检测的准确性较高,抗干扰能力较强,对入侵信息的准确检测概率提升效果显著。     

基于时滞的H∞滤波器设计及其在网络中的应用

本文讨论了基于时滞的不确定线性系统的H∞滤波器设计及其在网络中的应用问题. 首先考虑时滞的上下界, 并在不忽略Lyapunov泛函导数中的任何一项的前提下, 利用自由权矩阵方法, 提出了一种新的滤波误差系统的H∞性能分析和滤波器设计方法. 而且, 滤波器的系数矩阵可以通过解一组线性矩阵不等式(LMIs)得到. 然后, 将该方法应用于考虑网络影响的具有参数不确定性的线性系统的滤波问题. 与已有文献的方法相比, 结果的保守性大 大降低. 最后给出的数值实例说明了本文所给方法的有效性和优越性.     

Reconstruction of hillslope and valley paleotopography by application of a geomorphic model

Many applications in geology require estimation of the depth and thickness of lithologic layers based on limited observations. The boundaries of such layers are typically estimated using Kriging or other estimation methods that produce smooth surfaces. In some cases, however, smooth surfaces may be inappropriate. A boundary that is formed by a preserved hillslope and valley paleotopography, in particular, is expected to exhibit drainage characteristics and inherent roughness that are not consistent with standard estimation methods. This paper discusses the generalization of a technique originally designed to interpolate fluvially eroded topography. The method incorporates a simple river basin evolution model to generate realistic topography and adjusts an erodability parameter in space to match observed elevations. The method is generalized to allow flow to enter from outside the interpolation region, which is a likely scenario when reconstructing paleotopography. The method is then applied to the lower boundary of the Tshirege Member of the Bandelier Tuff, which underlies Los Alamos National Laboratory and Bandelier National Monument in north–central New Mexico. The method produces surfaces with major valleys that are consistent with previous studies. The method is also applied in a framework that estimates the likelihood that any particular point within the interpolation region drains through a specified boundary. Although the surfaces vary between simulations, most portions of the interpolation domain drain through consistent boundaries.     

基于Bloom Filter和概率分发队列的P2P网络快速查找算法

无结构化P2P网络资源定位过程中的响应时间、查准率及覆盖率难以同时被优化。提出一种面向有向无环随机网络的基于Bloom Filter和概率分发队列的快速查找算法BFPDQ(Bloom Filter and Probabilistic Distribution Queue),它用Bloom Filter表达和传递节点命中资源信息及查找请求信息,计算新查询消息与历史查询消息Bloom Filter语义向量相似度,并应用底层网络路径性能信息指导上层转发决策。概率分发队列(Probabilistic Distribution Queue,PDQ)把传统walkers表示成为查找消息分发队列,查找请求者协调各分发队列的查找方向和深度,并融合各队列查找过程中得到的定位消息。仿真实验表明,BFPDQ算法在保持较少冗余信息的同时有效缩短了响应时间。     

Multiresolution analysis of ridges and valleys in grey-scale images

Two methods for identifying and analyzing the multiresolution behavior of ridges and valleys in grey-scale images are described. The first method uses the tools of differential geometry to focus on local image behavior. The resulting vertex curves mark the tops of ridges and bottoms of valleys in an image. The second method focuses on the global drainage patterns of rainfall on a terrain map. The resulting watershed boundaries also identify the tops of ridges and bottoms of valleys in an image. By following these two geometric representations through scale space, the authors build resolution hierarchies on ridges and valleys in the image that can be utilized for interactive image segmentation     

基于数据的改进回声状态网络在高炉煤气发生量预测中的应用

以钢铁企业高炉煤气系统这一复杂生产过程为背景, 针对高炉煤气发生量的预测问题, 提出一种基于数据的网络模型预测方法. 鉴于生产数据含噪高的特点, 采用经验模态分解将历史数据分解为若干独立的固有模态函数, 将小尺度函数经低通滤波器自适应去噪后, 再对数据重构以建立预测模型. 在建模过程中提出一种改进的回声状态网络, 通过奇异值分解求取网络输出权值, 克服了线性回归算法出现的病态问题, 提高了模型的预测精度. 现场实际数据预测结果表明所提出方法的有效性, 为制定煤气管网平衡调度方案提供科学的决策支持.     

Nonlinear Filtering Under Stochastic Communication Protocol with Unknown Scheduling Probability

In this paper, the UKF-type nonlinear filtering problem is investigated for general nonlinear systems under stochastic communication protocols (SCPs) with unknown scheduling probabilities. In order to avoid the data collision and alleviate the network communication burden, SCPs, allowed only one sensor node to send data via the shared network, are exploited to orchestrate the scheduling order of sensor nodes. Different from traditional assumptions with accurate statistics, the scheduling probability of the selected node is unknown, but lies in a reliable interval with known upper and lower bounds. Due to the unknown probabilities, the exact estimation error covariance is not available and hence its upper bound is derived with the help of adding zero terms and eigenvalues of positive definite matrices. Such an upper bound is dependent on known upper and lower bounds of the scheduling probabilities and further utilized to reasonably design the filter gain at each time instant. In light of the obtained covariance and the filter gain, an improved unscented transformation is developed to carry out the designed UKF-type nonlinear filter by improving traditional approximate mean and covariance. Furthermore, the impact of the uncertain size of unknown scheduling probabilities is thoroughly discussed. Finally, a numerical example is given to confirm the effectiveness of the proposed nonlinear filter.

     

A new hierarchical moving curve-fitting algorithm for filtering lidar data for automatic DTM generation

Recent advances in laser scanning hardware have allowed rapid generation of high-resolution digital terrain models (DTMs) for large areas. However, the automatic discrimination of ground and non-ground light detection and ranging (lidar) points in areas covered by densely packed buildings or dense vegetation is difficult. In this paper, we introduce a new hierarchical moving curve-fitting filter algorithm that is designed to automatically and rapidly filter lidar data to permit automatic DTM generation. This algorithm is based on fitting a second-degree polynomial surface using flexible tiles of moving blocks and an adaptive threshold. The initial tile size is determined by the size of the largest building in the study area. Based on an adaptive threshold, non-ground points and ground points are classified and labelled step by step. In addition, we used a multi-scale weighted interpolation method to estimate the bare-earth elevation for non-ground points and obtain a recovered terrain model. Our experiments in four study areas showed that the new filtering method can separate ground and non-ground points in both urban areas and those covered by dense vegetation. The filter error ranged from 4.08% to 9.40% for Type I errors, from 2.48% to 7.63% for Type II errors, and from 5.01% to 7.40% for total errors. These errors are lower than those of triangulated irregular network filter algorithms.     

称重传感器非线性误差自适应补偿方法

额定量程内称重传感器的非线性误差不同,为此阐述了称重传感器的非线性误差特性,提出了一种非线性误差自适应分段补偿方法:在额定量程的上限区,采用基于径向基函数神经网络(RBFNN)的补偿网络完成传感器非线性误差补偿;在下限区,采用数字滤波器完成非线性误差补偿;在中间区,传感器不补偿。同时利用自适应选择网络,完成了分段补偿的选择。实验表明,采用这种方法补偿后的称重传感器下限区、中间区与上限区的最大相对误差分别由补偿前的0.2、0.4、1.37下降到0.16、0.04、0.07,补偿效果明显。     

FuzzyShell: a large-scale expert system shell using fuzzy logic foruncertainty reasoning

There exist in the literature today many contributions dealing with the incorporation of fuzzy logic in expert systems. However, unfortunately, much of what has been proposed can only be applied to small-scale expert systems; that is, when the number of rules is in the dozens as opposed to in the hundreds. The more traditional (nonfuzzy) expert systems are able to cope with large numbers of rules by using Rete networks for maintaining matches of all the rules and all the facts. (A Rete network obviates the need to match the rules with the facts on every cycle of the inference engine.) In this paper, we present a more general Rete network that is particularly suitable for reasoning with fuzzy logic. The generalized Rete network consists of a cascade of three networks: the pattern network, the join network, and the evidence aggregation network. The first two layers are modified versions of similar layers for the traditional Rete networks and the last, the aggregation layer, is a new concept that allows fuzzy evidence to be aggregated when fuzzy inferences are made about the same fuzzy variable by different rules     

Differentiation of Acacia koa forest stands across an elevation gradient in Hawai‘i using fine-resolution remotely sensed imagery

Koa (Acacia koa) forests are found across broad environmental gradients in the Hawaiian Islands. Previous studies have identified important environmental factors controlling stand structure and productivity at the plot level, but these have not been applied at the landscape level because of small-scale spatial variability. The goal of this study is to compare the differentiation of koa forest types across an elevation/temperature gradient ranging from 1200 to 2050 m asl (17–13°C mean annual temperature (MAT)) through the analysis of field measurements of forest structure and fine-resolution remotely sensed imagery. Several vegetation indices (VIs) (atmospherically resistant vegetation index (ARVI), enhanced vegetation index (EVI), normalized difference vegetation index (NDVI), soil-adjusted vegetation index (SAVI), modified soil-adjusted vegetation index (MSAVI), simple ratio (SR) and modified simple ratio (MSR)) are calculated from IKONOS satellite imagery of these stands and analysed using supervised classification techniques. This procedure allows a clear differentiation of koa stands from areas dominated by grasses, shrubs and bare lava. Across the elevation gradient, VIs allow differentiation of three koa forest stand classes at upper, intermediate and lower elevations. In agreement with the image classification, analysis of variance (ANOVA) of tree height and leaf phosphorus (P) suggests that there are also three significantly different groups of koa stands at those elevations. A landscape-scale map of land cover and koa stand classes demonstrates both the general trend with elevation and the small-scale heterogeneity that exists across the elevation gradient. Application of these classification techniques with fine spatial resolution imagery can improve the characterization of different koa stand types across the islands of Hawai‘i, which should aid both the conservation and utilization of this ecologically important species.     

Designing multilayer perceptrons from nearest-neighbor systems

Although multilayer perceptrons have been shown to be adept at providing good solutions to many problems, they have a major drawback in the very large amount of time needed for training (for example, on the order of CPU days for some of the author's experiments). The paper describes a method of producing a reasonable starting point by using a nearest-neighbor classifier. The method is further expanded to provide a method of ;programming' the upper layer of any network assuming the lower layers already exist.     

TecDEM: A MATLAB based toolbox for tectonic geomorphology, Part 2: Surface dynamics and basin analysis

We present the analytical capability of TecDEM, a MATLAB toolbox used in conjunction with Global DEMs for the extraction of tectonic geomorphologic information. TecDEM includes a suite of algorithms to analyze topography, extracted drainage networks and sub-basins. The aim of part 2 of this paper series is the generation of morphometric maps for surface dynamics and basin analysis. TecDEM therefore allows the extraction of parameters such as isobase, incision, drainage density and surface roughness maps. We also provide tools for basin asymmetry and hypsometric analysis. These are efficient graphical user interfaces (GUIs) for mapping drainage deviation from basin mid-line and basin hypsometry. A morphotectonic interpretation of the Kaghan Valley (Northern Pakistan) is performed with TecDEM and the findings indicate a high correlation between surface dynamics and basin analysis parameters with neotectonic features in the study area.     

Robust location tracking using a dual layer particle filter

Location awareness is an important part of many ubiquitous computing systems, but a perfect location system does not exist yet. Among many location tracking systems, we choose the radio frequency identification (RFID) system due to its various applications. However, the received signal strength indicator (RSSI) signals are too sensitive to the direction of the RFID reader’s antenna, the orientation of the RFID tag, human interference, and the diversity of propagation media that might be present. As a result, the direct use of a conventional particle filter does not provide satisfactory tracking performance. To overcome this problem, we suggest a dual layer particle filter, where the lower layer determines the tag’s location in the block level using a triangulation technique or the support vector machine (SVM) classifier, and the upper layer accurately estimates the tag’s location using the conventional particle filter within the pre-computed or classified block. This layered structure improves the location estimation and the tracking performance, because the location evidence from the lower layer effectively restricts the range of possible locations of the upper layer. We implement the proposed location tracking method using a ubiquitous RFID wireless network in an intelligent office, where several RFID readers are located in fixed locations and people or objects with active RFID tags move around the office. Extensive experiments show that the proposed location tracking method is so precise and robust that it is a good choice for person or object tracking in ubiquitous computing contexts. We also validate the usefulness of the proposed location tracking method by implementing it for a real-time people monitoring system in a noisy and complex steel mill.     

嵌入改进SENet的卷积神经网络连续血压预测

提出了基于改进SENet卷积神经网络和自学习参数滤波器的连续血压预测方法。实验结果表明,改进SENet可以有效增加简单卷积神经网络对时序数据的预测能力,在卷积层数为二层、三层和四层时比简单卷积神经网络预测精度提升了34.8%、23.5%和36.0%,在此基础上利用自学习参数滤波器消除血压预测波形中的毛刺,最终得到平滑的连续血压预测结果。     

Integrating automatically processed SPOT HRV Pan imagery in a DEM-based procedure for channel network extraction

The contribution of automatically processed Système Pour l'Observation de la Terre (SPOT) high resolution visible (HRV) Pan data as an ancillary source of information to a digital elevation model (DEM)-based method for channel network extraction is introduced. The image processing stage included an application of a Laplacian filter for edge detection. Edge pixels that were not contiguous with the main channel network were eliminated and the channel network was buffered and finally skeletonized to create channels with one-pixel width. A DEM-based approach was implemented for an overlapping area using the terrain analysis using digital elevation models (TauDEM) procedure based on the principles of the flow direction matrix method. The channel network that was extracted by the use of the two methods in fusion was tested against the imagery, and the DEM-based, channel networks and conformed to the reference data more accurately in terms of coverage of channels, network connectivity and location of extracted channels. A disadvantage of the data fusion is the additional, few, artificial channels.