Fall detection for multiple pedestrians using depth image processing technique

A fall detection method based on depth image analysis is proposed in this paper. As different from the conventional methods, if the pedestrians are partially overlapped or partially occluded, the proposed method is still able to detect fall events and has the following advantages: (1) single or multiple pedestrian detection; (2) recognition of human and non-human objects; (3) compensation for illumination, which is applicable in scenarios using indoor light sources of different colors; (4) using the central line of a human silhouette to obtain the pedestrian tilt angle; and (5) avoiding misrecognition of a squat or stoop as a fall. According to the experimental results, the precision of the proposed fall detection method is 94.31% and the recall is 85.57%. The proposed method is verified to be robust and specifically suitable for applying in family homes, corridors and other public places.     

基于内距离形状上下文的跌倒检测方法

在全球社会老龄化的大背景下,老年人的身体健康状况和晚年生活质量需要更多的关注。跌倒在老年人群中发生率高并且带来的后果比较严重。文中提出一种应用于家庭场景的基于Inner-Distance形状上下文（ Inner-Distance Shape Context,IDSC）的跌倒检测方法。该方法通过Inner-Distance形状上下文获得视频帧前景形状的描述信息,使用形状匹配方法对视频序列中人体形状变化进行量化。对形变量化信息使用动态时间规整（ Dynamic Time Warping,DTW）方法实现跌倒行为的判定。实验结果表明该方法可有效、快速地判断跌倒,提取结果具有较好的查准率和查全率。     

基于可见光-热红外视觉监控的车间工人跌倒检测算法

为了解决工厂车间视觉监控存在噪声干扰、光线变化、目标遮挡等问题,提出一种基于多模态视觉监控的工人跌倒检测算法.首先,采用热像仪和可见光相机获取车间内全天候监控图像,结合自适应滤波模型对图像进行降噪处理,以抑制环境噪声对监控图像的干扰;然后,构建一种改进的人体姿态特征提取网络,通过融合串联时间帧合并模块和位姿残差模块,以简化目标检测的特征图尺度,实现监控图像中工人区域被部分遮挡时姿态的实时、可靠预测;最后,设计人体轴线倾角、人体外接矩形框长宽比以及双膝盖点移动速度作为工人跌倒判别性特征,进而实现车间内工人的跌倒判别.在自建数据集和公开数据集上对所提出方法进行验证,实验结果表明,所提出算法的跌倒检测精度分别为95.6%和96.3%,与对比算法相比具有更好的准确性和实时性.     

基于黑板模式的人体检测系统设计与实现

为了实现广角镜视频图像中的人体检测,设计并实现了基于黑板模式的室内人体检测系统。该系统算法通过排除非人体目标实现人体检测。实验结果表明基于黑板模式的人体检测方法有效,即使在人体局部遮挡、头部旋转和皮肤颜色差异的情况下也同样适用。为使用广角镜摄像机进行室内人体检测提供了一种可行的方法。     

基于Kinect体感传感器的老年人跌倒自动检测

跌倒是独居老人最主要的意外风险之一,为快速有效获取跌倒信息,使老年人得到及时救助,提出一种基于Kinect体感传感器的人体跌倒自动检测方法,利用Kinect深度图像技术获取人体深度图像前景图,建立前景图三维包围盒,通过实时计算的三维包围盒的长、宽、高数值以及该数值的变化速度,判断人体跌倒是否发生。利用遮挡融合算法,解决了人体躯干被障碍物部分遮挡时,跌倒事件的检测和判定。在室内居家环境下进行了26种测试场景实验,检测误报率为2.0%~6.0%,漏报率为0~4.0%。该方法可以较为准确地实现人体跌倒自动检测。     

基于混合位姿估计模型的移动机器人三维地图创建方法

针对室内复杂环境下的稠密三维建模问题, 提出一种基于RGB-D 相机的移动机器人同时定位与三维地图创建方法. 该方法利用架设在移动机器人上的RGB-D 相机获取环境信息, 根据点云和纹理加权模型建立结合局部纹理约束的混合位姿估计方法, 确保定位精度的同时减小失败率. 在关键帧选取机制下, 结合视觉闭环检测方法, 运用树结构网络优化(TORO) 算法最小化闭环误差, 实现三维地图的全局一致性优化. 在室内环境下的实验结果验证了所提出算法的有效性和可行性.

     

结合边缘和区域的活动轮廓模型SAR图像目标轮廓提取

目的 目标轮廓表征了目标形状,可用于目标方位角估计、自动目标识别等,因此提取合成孔径雷达（SAR）图像中的目标轮廓受到了人们的广泛关注。受SAR图像乘性噪声的影响,传统的目标轮廓提取方法应用在SAR图像时失效。针对这一问题,提出一种将基于边缘的活动轮廓模型和基于区域的活动轮廓模型相结合的活动轮廓模型。方法 以真实SAR图像为基础,分析了向量场卷积（VFC）活动轮廓模型以及区域竞争（RC）活动轮廓模型各自的特点和优势,发现这两个模型存在一定的互补性,因此将这两个模型进行了结合,得到了一种新的SAR图像目标轮廓提取方法。结果 基于真实SAR图像的实验结果表明,本文方法能较好地应对SAR图像信噪比较低、目标边缘模糊等特点,能准确地获得SAR图像目标轮廓。结论 本文方法可用于执行实际的SAR图像轮廓提取任务,为后续的SAR图像自动识别和特征级图像融合等任务提供了较为优良的输入信息。     

基于视频和人体姿态估计的老年人摔倒监测研究

人口老龄化所带来的养老服务问题是现代社会面临的严重问题。例如在很多国家跌倒是造成老年人因伤致死的最大原因,因此如何对老年人进行自动摔倒监测就成为养老服务亟待解决的问题。目前,在室内摔倒监测领域中,基于可穿戴设备和基于环境传感器等主流摔倒监测方法面临着设备复杂、成本较高等问题。鉴于此,将人体姿态估计引入摔倒监测领域,提出了一种基于2D视频的摔倒监测算法。首先利用OpenPose数据集提取原始数据中人体关节的位置;其次利用这些具有增强特征的数据构建静态分类模型和动态分类模型;最后,在3个公共摔倒数据集上进行模型训练和摔倒监测的测试,取得了较好的效果,可以为摔倒监测相关研究提供一定的参考。     

一种基于 CNN 和人体椭圆轮廓运动特征的 摔倒检测方法

为了解决传统的使用几何特征检测摔倒的方法的不稳定、难于区别一些相似的活 动等问题,提出了一种基于卷积神经网络(CNN)和人体椭圆轮廓的运动特征的摔倒检测方法。 首先,使用高斯混合模型检测出人体目标并求出其最小外接椭圆轮廓。然后在每一帧的椭圆 轮廓中提取出长短轴之比、方向角和人体质心的竖直方向速度,融合成一个基于时间序列的 运动特征。最后,经过一个浅层的 CNN 对这些运动特征进行训练,用于摔倒判断,并区分相 似的活动。实验结果表明,本文方法和现有的方法相比,克服了几何特征的不稳定性,提高 了检测率。     

Automatic segmentation of liver & lesion detection using H-minima transform and connecting component labeling

Automatic segmentation of the liver and the Lesion detection can be a very challenging task due to its variability in size, shape, position and the presence of other organs with similar intensities. Manual segmentation and detection of a tumor is a time-consuming task and greatly depends upon the expertise and experience of the physician. We proposed a method which consists of automatic segmentation and detection of liver and lesion using CT scan modality. H-minima transform filter, Otsu global thresholds, Morphological opening by reconstruction and modified Connected Component Labeling algorithms are applied for liver segmentation. To keep the technique simple and effective, an appropriate range of threshold values are defined to detect different types of lesions. Performance of the proposed system is evaluated and compared with the state-of-the art algorithms. The results of the comparison show that the proposed approach is robust and efficient due to its simplicity. The dice coefficient score for the hepatic segmentation is 94% while sensitivity and specificity for hepatic lesion are 93% and 87% respectively.

     

Fall detection method based on Spatio-temporal feature fusion using combined two-channel classification

Nowadays, the growing population of senior citizens is a challenge for almost all developing countries. New technologies can help monitor elderlies at home by providing an innovative and secure environment and further enhancing their quality of living. Vision-based systems offer promising results in analyzing human posture and detecting abnormal events like falls. Falls appear to possess the most considerable risk for seniors living alone. In this article, a new fall detection method is proposed based on a fusion of motion-based and human shape-based features. Motion History Images (MHI) represent the temporal feature in our approach. Simultaneously, the height-to-width ratio and centroid of the moving person represent the spatial features. A two-channel classification model is designed using a threshold-based and a keyframe-based approach. The two channels are further combined based on any classification disparity for which more information is used to classify between falls and daily activities. Keyframes are selected based on the displacement of the spatial features having a threshold higher than a preset value. Keyframes are subject to a K-NN classification. The proposed algorithm delivers promising results on the UR fall detection dataset’s simulated fall and daily activity sequences. It provides satisfactory performance compared to existing state-of-the-art methods and shows a peak accuracy of 98.6% and recall of 100% in detecting falls. Specificity and precision are over 96%.

     

Automatic Feature Point Detection and Tracking of Human Actions in Time-of-flight Videos

Detecting feature points on the human body in video frames is a key step for tracking human movements. There have been methods developed that leverage models of human pose and classification of pixels of the body image. Yet, occlusion and robustness are still open challenges. In this paper, we present an automatic, model-free feature point detection and action tracking method using a time-of-flight camera. Our method automatically detects feature points for movement abstraction. To overcome errors caused by miss-detection and occlusion, a refinement method is devised that uses the trajectory of the feature points to correct the erroneous detections. Experiments were conducted using videos acquired with a Microsoft Kinect camera and a publicly available video set and comparisons were conducted with the state-of-the-art methods. The results demonstrated that our proposed method delivered improved and reliable performance with an average accuracy in the range of 90%. The trajectorybased refinement also demonstrated satisfactory effectiveness that recovers the detection with a success rate of 93.7%. Our method processed a frame in an average time of 71.1 ms.      

3D measures exploitation for a monocular semi-supervised fall detection system

Falls have been reported as the leading cause of injury-related visits to emergency departments and the primary etiology of accidental deaths in elderly. Thus, the development of robust home surveillance systems is of great importance. In this article, such a system is presented, which tries to address the fall detection problem through visual cues. The proposed methodology utilizes a fast, real-time background subtraction algorithm, based on motion information in the scene and pixels intensity, capable to operate properly in dynamically changing visual conditions, in order to detect the foreground object. At the same time, it exploits 3D space’s measures, through automatic camera calibration, to increase the robustness of fall detection algorithm which is based on semi-supervised learning approach. The above system uses a single monocular camera and is characterized by minimal computational cost and memory requirements that make it suitable for real-time large scale implementations.     

基于姿态估计的实时跌倒检测算法

为了快速、准确地检测老年人跌倒事件的发生,给出一种基于姿态估计的实时跌倒检测算法.首先利用基于深度学习的人体姿态估计算法得到人体关节点的坐标;然后,通过计算人体质心点的下降速度、跌倒后颈部关节点的纵坐标值是否大于阈值、以及肩部和腰部关节点的相对位置关系来判断跌倒是否发生.所给出的跌倒检测算法利用单目相机进行检测,便于以嵌入式方式应用于机器人.实验结果表明,所提出算法与当前先进方法相比取得了较好的效果.     

SAR图像舰船目标检测的信息几何方法

目的 舰船目标检测是合成孔径雷达（SAR）图像在海事监测领域中的一项重要应用。由于海面微波散射的复杂性,SAR图像中海杂波分布具有非均匀性、非平稳性等特点,传统的基于恒虚警率（CFAR）的SAR图像舰船检测算法难以适应复杂多变的海杂波环境,无法实现实时有效的智能检测任务。鉴于此,本文提出了基于信息几何的SAR图像船舰目标检测方法,旨在分析统计流形及其在参数空间中的几何结构,探讨信息几何在SAR图像目标检测应用中的切入点,从新的角度提升该应用领域的理论与技术水平。方法 首先,运用威布尔分布族对SAR图像中的海杂波进行统计建模,利用最大似然方法估计SAR图像局部邻域像素的分布参数,并将不同参数下的统计分布作为威布尔流形上的不同点;其次,融合高斯分布的费歇耳度量来构造威布尔流形空间中概率分布之间的测度,实现目标与背景区域的差异性表征;最后,利用最大类间方差法,实现SAR图像舰船目标检测。结果 实验和分析表明,相比于传统的基于恒虚警率的检测算法,信息几何方法可以有效地区分舰船目标和海杂波背景,降低虚警率,实现舰船目标显著性表示与检测。结论 由于舰船目标的复杂后向散射特性,如何有效地表征这一差异,是统计类检测算法的关键所在。本文依据信息几何理论,将概率分布族的参数空间视为微分流形,在参数流形上构造合适的黎曼度量,对SAR图像中各像素局部邻域进行测度表征,可以显著性表示目标与背景杂波之间的统计差异,实现舰船目标检测。     

Complete 3D Foot Scanning System Using 360 Degree Rotational and Translational Laser Triangulation Sensors

This paper proposes a new type of 3D foot scanning system using rotational and translational 3D scanning stages. Commercial 3D foot scanning systems (or scanners) mostly employ the laser triangulation method and three or more linear stages to scan the entire 3D shape of the foot. We introduce a new foot scanning method using only two laser-camera triangulation sensors. The proposed scanning system consists of a 360° rotational and a linear translational 3D sensors. The rotational sensor employs two line lasers with a vision camera to solve an occlusion problem of the rotational stage and acquires the 3D shape of the upper part of the foot. The translational sensor consists of a line laser and a vision camera and acquires the 3D shape of the foot sole. The performance of the proposed scanning technique is verified using plastic models and human feet. In average, about 0.5 mm reconstruction accuracy is obtained by the proposed technique.

     

融合运动特征和深度学习的跌倒检测算法

为了利用计算机视觉技术准确检测老年人的跌倒状况,针对现有跌倒检测算法中人为设计特征造成的不完备性以及跌倒检测过程中前后景分离困难、目标混淆、运动目标丢失、跌倒检测准确率低等问题,提出了一种融合人体运动信息的深度学习跌倒检测算法对人体跌倒状态进行检测。首先,通过改进YOLOv3网络进行前景与背景的分离,并根据YOLOv3网络的检测结果对前景人体目标进行最小外接矩形标记;其次,分析人体跌倒过程中的运动特征,将人体运动特征向量化并通过Sigmoid激活函数转化为0到1之间的运动权重信息;最后,通过全连接层将将运动特征与卷积神经网络（CNN）提取的特征进行拼接和融合从而实现人体跌倒分类判别。将所提跌倒检测算法与背景差分、高斯混合、VIBE、方向梯度直方图（HOG）等人体目标检测算法及阈值法、分级法、支持向量机（SVM）分类和CNN分类等人体跌倒判断方案进行了对比实验,并将所提跌倒检测算法在不同光照条件下和混合日常噪声运动干扰下进行了实验,结果表明所提算法在环境适应性和跌倒检测准确率上都优于传统的人体跌倒检测方法。该算法能有效检测出视频中的人体并对人体跌倒状态进行准确检测,进一步验证了融合运动信息的深度学习识别方法在视频跌倒行为分析上的可行性与高效性。     

Human fall detection in surveillance video based on PCANet

Fall incidents have been reported as the second most common cause of death, especially for elderly people. Human fall detection is necessary in smart home healthcare systems. Recently various fall detection approaches have been proposed., among which computer vision based approaches offer a promising and effective way. In this paper, we proposed a new framework for fall detection based on automatic feature learning methods. First, the extracted frames, including human from video sequences of different views, form the training set. Then, a PCANet model is trained by using all samples to predict the label of every frame. Because a fall behavior is contained in many continuous frames, the reliable fall detection should not only analyze one frame but also a video sequence. Based on the prediction result of the trained PCANet model for each frame, an action model is further obtained by SVM with the predicted labels of frames in video sequences. Experiments show that the proposed method achieved reliable results compared with other commonly used methods based on the multiple cameras fall dataset, and a better result is further achieved in our dataset which contains more training samples.     

Color-based skin segmentation in videos using a multi-step spatial method

Recently, skin detection has been employed in multifarious applications of computer vision including face detection, gesture recognition, etc. This is mainly due to the appealing characteristics of skin color and its potency to segment objects. However, there are certain challenges involved in utilizing human complexion as a feature to detect faces, and they have led to the inefficiency of many methods. In order to counteract these factors, in this paper, a skin segmentation method which exploits a multi step diffusion algorithm to detect skin regions is presented. The method starts with conservative extraction of skin seeds in each frame which is accomplished by using fusion of ternary-based human motion detection, modified Bayesian classifier, and a feedback mechanism. Subsequently, these candidate skin pixels are utilized in a 2-stage diffusion scheme to detect other skin pixels. Both quantitative and qualitative results demonstrate the effectiveness of the proposed system in comparison with other works.