基于非对称增强注意力与特征交叉融合的行人重识别方法

针对现有的行人重识别方法提取到的特征信息充分性与辨识性不足导致检索精度低的问题,提出一种基于非对称增强注意力与特征交叉融合的行人重识别方法。首先,构建非对称增强注意力模块,通过多重池化聚合的跨邻域通道交互注意力增强显著特征表示,使网络聚焦于图像中的行人区域;其次,考虑到网络各层特征间的差异性与关联性,构建特征交叉融合模块,利用交叉融合方式实现同层不同级特征的跨层级融合,进而实现多尺度融合;最后,水平切分输出特征以获取局部特征,从而实现在特定区域上描述行人。在Market1501、DukeMTMC-reID与CUHK03这3个公开数据集上对提出的方法进行了验证,首位命中率(Rank-1)分别达到了93.5%、85.1%和64.3%,证明了该方法在提升行人重识别性能上具有优越性。     

Multiple Pedestrian Detection and Tracking in Night Vision Surveillance Systems

Pedestrian detection and tracking are vital elements of today’s surveillance systems, which make daily life safe for humans. Thus, human detection and visualization have become essential inventions in the field of computer vision. Hence, developing a surveillance system with multiple object recognition and tracking, especially in low light and night-time, is still challenging. Therefore, we propose a novel system based on machine learning and image processing to provide an efficient surveillance system for pedestrian detection and tracking at night. In particular, we propose a system that tackles a two-fold problem by detecting multiple pedestrians in infrared (IR) images using machine learning and tracking them using particle filters. Moreover, a random forest classifier is adopted for image segmentation to identify pedestrians in an image. The result of detection is investigated by particle filter to solve pedestrian tracking. Through the extensive experiment, our system shows 93% segmentation accuracy using a random forest algorithm that demonstrates high accuracy for background and roof classes. Moreover, the system achieved a detection accuracy of 90% using multiple template matching techniques and 81% accuracy for pedestrian tracking. Furthermore, our system can identify that the detected object is a human. Hence, our system provided the best results compared to the state-of-art systems, which proves the effectiveness of the techniques used for image segmentation, classification, and tracking. The presented method is applicable for human detection/tracking, crowd analysis, and monitoring pedestrians in IR video surveillance.     

Centralized embedding hypersphere feature learning for person re-identification

ABSTRACT

Deep metric learning has become a general method for person re-identification (ReID) recently. Existing methods train ReID model with various loss functions to learn feature representation and identify pedestrian. However, the interaction between person features and classification vectors in the training process is rarely concerned. Distribution of pedestrian features will greatly affect convergence of the model and the pedestrian similarity computing in the test phase. In this paper, we formulate improved softmax function to learn pedestrian features and classification vectors. Our method applies pedestrian feature representation to be scattered across the coordinate space and embedding hypersphere to solve the classification problem. Then, we propose an end-to-end convolutional neural network (CNN) framework with improved softmax function to improve the performance of pedestrian features. Finally, experiments are performed on four challenging datasets. The results demonstrate that our work is competitive compared to the state-of-the-art.     

视频监控中基于在线多核学习的目标再现识别

在非重叠多摄像机或单摄像机视频监控中,识别跟踪目标的再次出现很重要.针对传统支持向量机方法在特征融合方面的缺陷,本文提出了一种新的基于在线多核学习的人体目标再现识别方法.该方法对跟踪目标视频前景图像序列提取具有互补性的视觉单词树直方图和全局颜色直方图二种特征,再采用多核学习方法在线训练人体目标视觉外观,从而得到多核特征融合模型.实验结果表明,该方法能快速训练人体目标外观模型,满足视频监控的实时要求,多核融合模型获得了比单一特征模型和单核支持向量机方法更高的识别性能.     

一种基于平均步态能量图的身份识别算法

提出一种基于步态能量图（GEI）的嵌入式隐马尔可夫模型（e-HMM）身份识别方法。首先通过预处理提取出运动人体的侧面轮廓,根据步态下肢的摆动距离统计出步态周期,得到平均步态能量图。对能量图的各区域进行分析,利用二维离散余弦变换（2D-DCT）将能量图观测块转化为观测向量,实现嵌入式隐马尔可夫模型的训练和身份识别。最后在USF和CASIA步态数据库上对所提出的算法进行实验。实验表明该方法具有较好的识别性能,是一种有效的步态识别方法。     

基于SVM的多生物特征融合识别算法

针对单生物特征识别的局限性,提出融合手背静脉和虹膜两种生物特征实现身份识别.基于尺度不变特征变换(SIFT)提取手背静脉的局部SIFT特征并对特征点进行匹配,利用特征匹配率作为手背静脉图像的相似度测度.通过Haar小波变换实现虹膜特征编码,利用加权汉明距对虹膜进行相似度测试.最后基于支持向量机(SVM)实现两种生物特征在匹配层的融合识别.利用CASIA虹膜数据库和TJU手背静脉数据库对算法性能进行测试,其等错率为0.02%,实验结果表明,该融合算法具有很高的识别性能,为生物特征识别研究提供了新思路.     

Pedestrian head translation,rotation and impact velocity: The influence of vehicle speed,pedestrian speed and pedestrian gait

In road traffic collisions, pedestrian injuries and fatalities account for approximately 11% and 20% of casualties in the USA and the EU, respectively. In many less motorised countries, the majority of victims are pedestrians. The significant influences of vehicle speed, pedestrian speed and pedestrian gait on pedestrian post-impact kinematics have been qualitatively noted in the literature, but there has been no quantitative approach to this problem. In this paper, the MADYMO MultiBody (MB) pedestrian model is used to analyse the influences of vehicle speed, pedestrian speed and pedestrian gait on the transverse translation of the pedestrian's head, head rotation about the vertical head axis and head impact velocity. Transverse translation has implications for injury severity because of variations in local vehicle stiffness. Head rotation is related to pedestrian stance at impact, which is known to affect the kinematics of a collision. Increased head impact velocity results in greater head injury severity. The results show that transverse translation of the head relative to the primary contact location of the pedestrian on the vehicle decreases with increasing vehicle speed and increases linearly with increasing pedestrian speed. Head rotation decreases with increasing vehicle speed and increases linearly with increasing pedestrian speed, but these variations are small. The range of head rotation values decreases with increasing vehicle speed. Head impact velocity increases linearly with vehicle speed and is largely independent of pedestrian speed. Transverse translation, head rotation and head impact velocity all vary cyclically with gait in clearly definable patterns.     

Towards an integrated approach of pedestrian behaviour and exposure

In this paper, an integrated methodology for the analysis of pedestrian behaviour and exposure is proposed, allowing to identify and quantify the effect of pedestrian behaviour, road and traffic characteristics on pedestrian risk exposure, for each pedestrian and for populations of pedestrians. The paper builds on existing research on pedestrian exposure, namely the Routledge microscopic indicator, proposes adjustments to take into account road, traffic and human factors and extends the use of this indicator on area-wide level. Moreover, this paper uses integrated choice and latent variables (ICLV) models of pedestrian behaviour, taking into account road, traffic and human factors. Finally, a methodology is proposed for the integrated estimation of pedestrian behaviour and exposure on the basis of road, traffic and human factors. The method is tested with data from a field survey in Athens, Greece, which used pedestrian behaviour observations as well as a questionnaire on human factors of pedestrian behaviour. The data were used (i) to develop ICLV models of pedestrian behaviour and (ii) to estimate the behaviour and exposure of pedestrians for different road, traffic and behavioural scenarios. The results suggest that both pedestrian behaviour and exposure are largely defined by a small number of factors: road type, traffic volume and pedestrian risk-taking. The probability for risk-taking behaviour and the related exposure decrease in less demanding road and traffic environments. A synthesis of the results allows to enhance the understanding of the interactions between behaviour and exposure of pedestrians and to identify conditions of increased risk exposure. These conditions include principal urban arterials (where risk-taking behaviour is low but the related exposure is very high) and minor arterials (where risk-taking behaviour is more frequent, and the related exposure is still high). A “paradox” of increased risk-taking behaviour of pedestrians with low exposure is found, suggesting that these pedestrians may partly compensate in moderate traffic conditions due to their increased walking speed.     

An Efficient Crossing-Line Crowd Counting Algorithm with Two-Stage Detection

Crowd counting is a challenging task in crowded scenes due to heavy occlusions, appearance variations and perspective distortions. Current crowd counting methods typically operate on an image patch level with overlaps, then sum over the patches to get the final count. In this paper we describe a real-time pedestrian counting framework based on a two-stage human detection algorithm. Existing works with overhead cameras is mainly based on visual tracking, and their robustness is rather limited. On the other hand, some works, which focus on improving the performances, are too complicated to be realistic. By adopting a line sampling process, a temporal slice image can be obtained for pedestrian counting without the need for visual tracking. Only ten low level features are extracted from the input image to establish a feature vector. As a result, our algorithm is more efficient and accurate than existing methods. Pedestrians in the temporal slice image are then located by the two-stage detection algorithm, which is largely based on support vector machine and affinity propagation clustering. Moreover, a novel algorithm is proposed to determine the moving directions of pedestrians by comparing the centers of them in two temporal slice images. Extensive experiments reveal that our system achieves satisfaction performances in terms of both robustness and efficiency.     

A Multi-Feature Learning Model with Enhanced Local Attention for Vehicle Re-Identification

Vehicle re-identification (ReID) aims to retrieve the target vehicle in an extensive image gallery through its appearances from various views in the cross-camera scenario. It has gradually become a core technology of intelligent transportation system. Most existing vehicle re-identification models adopt the joint learning of global and local features. However, they directly use the extracted global features, resulting in insufficient feature expression. Moreover, local features are primarily obtained through advanced annotation and complex attention mechanisms, which require additional costs. To solve this issue, a multi-feature learning model with enhanced local attention for vehicle re-identification (MFELA) is proposed in this paper. The model consists of global and local branches. The global branch utilizes both middle and high-level semantic features of ResNet50 to enhance the global representation capability. In addition, multi-scale pooling operations are used to obtain multi-scale information. While the local branch utilizes the proposed Region Batch Dropblock (RBD), which encourages the model to learn discriminative features for different local regions and simultaneously drops corresponding same areas randomly in a batch during training to enhance the attention to local regions. Then features from both branches are combined to provide a more comprehensive and distinctive feature representation. Extensive experiments on VeRi-776 and VehicleID datasets prove that our method has excellent performance.     

Medical image fusion based on nuclear norm minimization

Medical image fusion plays an important role in diagnosis and treatment of diseases such as image‐guided radiotherapy and surgery. Although numerous medical image fusion methods have been proposed, most approaches have not touched the low rank nature of matrix formed by medical image, which usually lead to fusion image distortion and image information loss. These methods also often lack universality when dealing with different kinds of medical images. In this article, we propose a novel medical image fusion to overcome aforementioned issues on existing methods with the aid of low rank matrix approximation with nuclear norm minimization (NNM) constraint. The workflow of our method is described as: firstly, nonlocal similar patches across the medical image are searched by block matching for local patch in source images. Second, a fused matrix is stacking by shared nonlocal similarity patches, then the low rank matrix approximation methods under nuclear norm minimization can be used to recover low rank feature of fused matrix. Finally, fused image can be gotten by aggregating all the fused patches. Experimental results show that the proposed method is superior to other methods in both subjectively visual performance and objective criteria. © 2015 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 25, 310–316, 2015     

Pedestrian signalization and the risk of pedestrian-motor vehicle collisions in Lima,Peru

Safe walking environments are essential for protecting pedestrians and promoting physical activity. In Peru, pedestrians comprise over three-quarters of road fatality victims. Pedestrian signalization plays an important role managing pedestrian and vehicle traffic and may help improve pedestrian safety. We examined the relationship between pedestrian-motor vehicle collisions and the presence of visible traffic signals, pedestrian signals, and signal timing to determine whether these countermeasures improved pedestrian safety. A matched case-control design was used where the units of study were crossing locations. We randomly sampled 97 control-matched collisions (weighted N = 1134) at intersections occurring from October, 2010 to January, 2011 in Lima. Each case-control pair was matched on proximity, street classification, and number of lanes. Sites were visited between February, 2011 and September, 2011. Each analysis accounted for sampling weight and matching and was adjusted for vehicle and pedestrian traffic flow, crossing width, and mean vehicle speed. Collisions were more common where a phased pedestrian signal (green or red-light signal) was present compared to no signalization (odds ratio [OR] 8.88, 95% Confidence Interval [CI] 1.32–59.6). A longer pedestrian-specific signal duration was associated with collision risk (OR 5.31, 95% CI 1.02–9.60 per 15-s interval). Collisions occurred more commonly in the presence of any signalization visible to pedestrians or pedestrian-specific signalization, though these associations were not statistically significant. Signalization efforts were not associated with lower risk for pedestrians; rather, they were associated with an increased risk of pedestrian-vehicle collisions.     

Attention-Aware Network with Latent Semantic Analysis for Clothing Invariant Gait Recognition

Gait recognition is a complicated task due to the existence of co-factors like carrying conditions, clothing, viewpoints, and surfaces which change the appearance of gait more or less. Among those co-factors, clothing analysis is the most challenging one in the area. Conventional methods which are proposed for clothing invariant gait recognition show the body parts and the underlying relationships from them are important for gait recognition. Fortunately, attention mechanism shows dramatic performance for highlighting discriminative regions. Meanwhile, latent semantic analysis is known for the ability of capturing latent semantic variables to represent the underlying attributes and capturing the relationships from the raw input. Thus, we propose a new CNN-based method which leverages advantage of the latent semantic analysis and attention mechanism. Based on discriminative features extracted using attention and the latent semantic analysis module respectively, multi-modal fusion method is proposed to fuse those features for its high fault tolerance in the decision level. Experiments on the most challenging clothing variation dataset: OU-ISIR TEADMILL dataset B show that our method outperforms other state-of-art gait approaches.     

Exploring factors affecting pedestrians’ red-light running behaviors at intersections in China

Pedestrians’ Red-light running behavior is one of the most critical factors for pedestrian involved traffic crashes at intersections in China. The primary objective of this study is to explore how various factors affect pedestrians’ red-light running behaviors at intersection areas, using the data collected from Hefei, China. A questionnaire was well designed aiming at collecting pedestrians’ socio-economic characteristics, trip related features, and attribute variables in different crossing facilities. Based on 631 valid samples, a binomial logistic model was established to evaluate the impacts of contributing factors on pedestrians’ red-light running behavior. The modeling results show that four variables significantly affect the probability of pedestrians’ red-light running behavior, which are the trip purpose, time period in one day, pedestrian’s attitude towards whether to run a red light when in hurry, and pedestrian’s attitude towards whether quality of road facility affects crossing behavior. With those variables, the probability of pedestrians’ red-light running behavior at intersections could be predicted. Findings of this study can help understand why pedestrians in China run red-lights and identify which pedestrian groups and intersections are more likely to have such behaviors. This study can also help propose countermeasures more efficiently to reduce pedestrian-related crashes at intersections in China.     

Medical image fusion based on laws of texture energy measures in stationary wavelet transform domain

Medical image fusion is widely used in various clinical procedures for the precise diagnosis of a disease. Image fusion procedures are used to assist real-time image-guided surgery. These procedures demand more accuracy and less computational complexity in modern diagnostics. Through the present work, we proposed a novel image fusion method based on stationary wavelet transform (SWT) and texture energy measures (TEMs) to address poor contrast and high-computational complexity issues of fusion outcomes. SWT extracts approximate and detail information of source images. TEMs have the capability to capture various features of the image. These are considered for fusion of approximate information. In addition, the morphological operations are used to refine the fusion process. Datasets consisting of images of seven patients suffering from neurological disorders are used in this study. Quantitative comparison of fusion results with visual information fidelity-based image fusion quality metric, ratio of spatial frequency error, edge information-based image fusion quality metric, and structural similarity index-based image fusion quality metrics proved the superiority. Also, the proposed method is superior in terms of average execution time to state-of-the-art image fusion methods. The proposed work can be extended for fusion of other imaging modalities like fusion of functional image with an anatomical image. Suitability of the fused images by the proposed method for image analysis tasks needs to be studied.     

Impact of grade separator on pedestrian risk taking behavior

Pedestrians on Delhi roads are often exposed to high risks. This is because the basic needs of pedestrians are not recognized as a part of the urban transport infrastructure improvement projects in Delhi. Rather, an ever increasing number of cars and motorized two-wheelers encourage the construction of large numbers of flyovers/grade separators to facilitate signal free movement for motorized vehicles, exposing pedestrians to greater risk. This paper describes the statistical analysis of pedestrian risk taking behavior while crossing the road, before and after the construction of a grade separator at an intersection of Delhi. A significant number of pedestrians are willing to take risks in both before and after situations. The results indicate that absence of signals make pedestrians behave independently, leading to increased variability in their risk taking behavior. Variability in the speeds of all categories of vehicles has increased after the construction of grade separators. After the construction of the grade separator, the waiting time of pedestrians at the starting point of crossing has increased and the correlation between waiting times and gaps accepted by pedestrians show that after certain time of waiting, pedestrians become impatient and accepts smaller gap size to cross the road. A Logistic regression model is fitted by assuming that the probability of road crossing by pedestrians depends on the gap size (in s) between pedestrian and conflicting vehicles, sex, age, type of pedestrians (single or in a group) and type of conflicting vehicles. The results of Logistic regression explained that before the construction of the grade separator the probability of road crossing by the pedestrian depends on only the gap size parameter; however after the construction of the grade separator, other parameters become significant in determining pedestrian risk taking behavior.     

利用小波边缘增强的可靠性匹配方法

为提高景像匹配的可靠性,提出了一种基于小波边缘增强的可靠性匹配方法。采用四阶中心B样条小波对实时图和基准图进行小波分解和边缘增强,以提取可靠的边缘特征。在粗尺度上基于小波边缘增强图进行相关匹配,选择相似度曲面上前5个峰值点作为候选匹配点,保证正确匹配点可以包含在候选匹配点中。对实时图进行旋转校正,利用候选匹配点处的局部灰度特征确定边缘提取的双阈值,应用形态学连接算子来获得二值边缘图,再进行相关匹配,筛选出正确匹配点。在细尺度上获得精确的定位点。应用该方法进行匹配实验,其匹配概率比基于小波二值边缘提取的匹配算法提高了6.95%。     

考虑eHMI的行人过街行为研究与决策模型构建

目的 针对行人与自动驾驶汽车的交互过程,从行人的角度出发,探索性地提出行人在自动驾驶汽车前的过街行为决策模型。方法 首先,将行人过街情景进行分析定义,针对行人的过街意向和与自动驾驶汽车沟通意愿总结出两个典型场景;然后,利用潜变量分析方法将行人在自动驾驶汽车前过街行为的影响因素进行降维分类,并对车外人机交互界面进行设计定义,构建各潜变量的影响因子与测量量表;最后,运用有序Logistic回归方法分析各影响因素对行人过街意向和沟通意愿的影响,构建行人在自动驾驶汽车前的过街行为决策模型。结果 量化分析行人过街行为影响因素与其过街决策间相关性及内在关系,提出过程中决定性人车交互方式及各影响因素的变化过程。结论 研究提出考虑eHMI的行人过街行为决策模型,将行人过街决策过程分为三个阶段,并总结出车辆行为线索和eHMI线索作用过程的决定性变化曲线及行人与自动驾驶汽车交互流程与关键性节点。