融合时域信息的修正分水岭视频分割新方法

提出一种将时域信息融入分水岭的视频分割新方法,以帧间变化检测为基础,通过运动边缘信息得到对象的初始模型,利用时域信息得到前景和背景的标识,结合提出的彩色多尺度形态学梯度算子进行分水岭分割,得到具有精确边界的视频对象,对慢变和快变的目标均有良好的效果,能够检测新出现的运动对象和现有对象的消失,能够定位和跟踪运动目标.继承了变化检测和分水岭算法速度快的优点,克服了两者易受噪声影响的缺点.     

基于RGB-D融合的密集遮挡抓取检测

针对当前抓取检测模型对密集遮挡物体的检测效果差以及人工数据标注工作量大的问题,提出基于RGB-D图像融合的目标检测与抓取检测分步骤进行的改进方案.新方案支持将单物体图像训练的抓取检测模型直接应用于密集遮挡的多物体图像场景中.首先,考虑到密集遮挡场景下抓取物具有多尺度的特点,提出子阶段路径聚合(SPA)的多尺度特征融合模块,用于丰富RGB-D特征级别融合的目标检测模型SPA-YOLO-Fusion的高维语义特征信息,以便于检测模型定位所有的抓取物;其次,使用基于RGB-D像素级别融合的GR-ConvNet抓取检测模型估计每个物体的抓取点,并提出背景填充的图像预处理算法来降低密集遮挡物体的相互影响;最后,使用机械臂对目标点进行抓取.在LineMOD数据集上对目标检测模型进行测试,实验结果表明SPA-YOLO-Fusion的mAP比YOLOv3-tiny与YOLOv4-tiny分别提高了10%与7%.从实际场景中采集图像制作YODO_Grasp抓取检测数据集并进行测试,结果表明增加背景填充预处理算法的GR-ConvNet的抓取检测精度比原模型提高了23%.     

人脸检测中眼睛精确定位的研究

人脸检测在许多应用中都是重要的一个处理阶段,例如人脸识别、电视会议、人机界面等。眼睛是一个在人脸检测中极为重要的人脸特征,因此一种快速可靠的精确定位眼睛的算法对许多实际的应用是十分重要的。该文提出了一种新颖的精确定位眼睛的方法,该方法由两部分组成:第一部分,通过人脸区域分割、五官定位、人脸确认三个步骤在复杂背景中进行人脸区域的检测;第二部分,在检测到人脸区域和眼睛大致位置的基础上,用一种新的椭圆检测算法精确定位眼睛虹膜的位置。实验证明该文所提出的算法是快速可靠的。     

Moving Objects Localization by Local Regions Based Level Set: Application on Urban Traffic

In this paper, a novel method for locating multiple moving objects in a video sequences captured by a stationary camera is proposed. In order to determine the precise location of the objects in an image, a new local regions based level set model is carried out. The whole process consists of two main parts: the global detection and the fine localization. During the global detection, the presence or absence of an object in an image is determined by the Mixture of Gaussians method. For the fine localization, we propose to reformulate global energies by utilizing little squared windows centered on each point over a thin band surrounding the zero level set, hence the object contour can be reshaped into small local interior and exterior regions that lead to compute a family of adaptive local energies, which enables us to well localize the moving objects. Moreover, we propose to adapt the smoothness of the contours, and the accuracy of the objects’ perimeter of different shapes with an automatic stopping criterion. The proposed method has been tested on different real urban traffic videos, and the experiment results demonstrate that our algorithm can locate effectively and accurately the moving objects; optimize the results of the localized objects and also decrease the computations load.     

基于非对称跨模态融合的RGB-D显著目标检测

目前大多数RGB-D显著目标检测方法在RGB特征和Depth特征的融合过程中采用对称结构,对两种特征进行相同的操作,忽视了RGB图像和Depth图像的差异性,易造成错误的检测结果.针对该问题,提出一种基于非对称结构的跨模态融合RGB-D显著目标检测方法,利用全局感知模块提取RGB图像的全局特征,并设计了深度去噪模块滤除低质量Depth图像中的大量噪声;再通过所提出的非对称融合模块,充分利用两种特征间的差异性,使用Depth特征定位显著目标,用于指导RGB特征融合,补足显著目标的细节信息,利用两种特征各自的优势形成互补.通过在4个公开的RGB-D显著目标检测数据集上进行大量实验,验证所提出的方法优于当前的主流方法.     

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。     

基于隐含多项式曲线的物体描述与对称性检测

物体的对称性检测是图像分析，计算视觉中一个重要研究课题，在形状匹配，基于模型的目标匹配，三维物体的重建，图像压缩和图像数据库检索等许多领域都有重要应用，但是，对称性检测一直是计算机视觉的难题，至今缺少通用性好，能同时检测各种对称性并且效率较高的算法，利用隐含多项式曲线对复杂物体的良好描述能力，提出了基于隐含多项式曲一的物体对称性检测方法，这将平面图形的对称性检测问题转化对隐含多项式曲线的研究，理论与实验表明该算法能检测物体的各种对称性，算法简单，而且还可以检测出有噪音和缺失部分信息的物体对称性     

自动分割视频序列中运动物体的新算法：多特征联合方法

视频编码标准MPEG-4增加了适于多种应用的基于视频内容的功能,为了支持这一功能和提高编码效率,MPEG-4将视频序列中的每一帧分解成视频对象面(VOP);另外,由于基于内容的视频检索和视频监控系统均期望用分割出的关键视频对象紧致地表示一个序列,同时由于视频分割技术在模式识别、计算机视觉等领域也得到了广泛的应用,因此,分割视频运动物体并跟踪运动物体的变化变得至关重要.为了对视频中运动物体进行有效的分割,在帧差图象的基础上,采用Canny边缘检测和随机信号的高阶矩检测相结合的方法,来自动分割视频序列的前景区域和背景区域,并在前景区域中应用区域生长法进行颜色分割,以精确提取运动物体的边缘;还利用边缘和颜色特征来对分割出的运动物体建立模板,用于解决非刚体运动中局部暂时停止运动的情况.实验结果表明,此方法可以有效地分割运动物体,并能跟踪运动物体的变化.     

一种室内动态场景下RGB-D SLAM的前端优化方法

针对动态物体会导致位姿估计误差过大的问题,提出一种适用于动态场景下的RGB-D SLAM算法。首先,将目标检测的结果根据物体是否具有自主移动能力划分为动态物体与潜在动态物体,并使用大津法在对应的深度图中完成检测框内区域的前、背景分割;然后,结合检测框的位置坐标与深度信息确定潜在动态物体与动态物体在三维空间内是否关联,初步筛选出具有一定运动概率的潜在动态物体;最后,通过运动一致性检测算法来最终确定是否将潜在动态物体作为动态物体处理。实验部分选用公开的TUM数据集,结果表明,该算法在保持实时性的同时具有较高的定位精度和鲁棒性。     

基于语义分割动态特征点剔除的SLAM算法

针对动态物体容易干扰SLAM建图准确性的问题,提出了一种新的动态环境下的RGB-D SLAM框架,将深度学习中的神经网络与运动信息相结合。首先,算法使用Mask R-CNN网络检测可能生成动态对象掩模的潜在运动对象。其次,算法将光流方法和Mask R-CNN相结合进行全动态特征点的剔除。最后在TUM RGB-D数据集下的实验结果表明,该方法可以提高SLAM系统在动态环境下的位姿估计精度,比现有的ORB-SLAM2的表现效果更好。     

Recent advances in 3D object detection based on RGB-D: A survey

3D object detection is a critical part of environmental perception systems and one of the most fundamental tasks in understanding the 3D visual world, which benefit a series of downstream real-world applications. RGB-D images include object texture and semantic information, as well as depth information describing spatial geometry. Recently, numerous 3D object detection models for RGB-D images have been proposed with excellent performance, but summaries in this area are still absent. To stimulate future research, this paper provides a detailed analysis of current developments in 3D object detection methods for RGB-D images to motivate future research. It covers three major parts, including background on 3D object detection, RGB-D data details, and comparative results of state-of-the-art methods on several publicly available datasets, with an emphasis on contributions, design ideas, and limitations, as well as insightful observations and inspiring future research directions.     

Computing multiscale curve and surface skeletons of genus 0 shapes using a global importance measure

We present a practical algorithm for computing robust, multiscale curve and surface skeletons of 3D objects. Based on a model which follows an advection principle, we assign to each point on the skeleton a part of the object surface, called the collapse. The size of the collapse is used as a uniform importance measure for the curve and surface skeleton, so that both can be simplified by imposing a single threshold on this intuitive measure. The simplified skeletons are connected by default, without special precautions, due to the monotonicity of the importance measure. The skeletons possess additional desirable properties: They are centered, robust to noise, hierarchical, and provide a natural skeleton-to-boundary mapping. We present a voxel-based algorithm that is straightforward to implement and simple to use. We illustrate our method on several realistic 3D objects.     

Consistent mesh partitioning and skeletonisation using the shape diameter function

Mesh partitioning and skeletonisation are fundamental for many computer graphics and animation techniques. Because of the close link between an object’s skeleton and its boundary, these two problems are in many cases complementary. Any partitioning of the object can assist in the creation of a skeleton and any segmentation of the skeleton can infer a partitioning of the object. In this paper, we consider these two problems on a wide variety of meshes, and strive to construct partitioning and skeletons which remain consistent across a family of objects, not a single one. Such families can consist of either a single object in multiple poses and resolutions, or multiple objects which have a general common shape. To achieve consistency, we base our algorithms on a volume-based shape-function called the shape-diameter-function (SDF), which remains largely oblivious to pose changes of the same object and maintains similar values in analogue parts of different objects. The SDF is a scalar function defined on the mesh surface; however, it expresses a measure of the diameter of the object’s volume in the neighborhood of each point on the surface. Using the SDF we are able to process and manipulate families of objects which contain similarities using a simple and consistent algorithm: consistently partitioning and creating skeletons among multiple meshes.     

A topology-preserving parallel 3D thinning algorithm for extracting the curve skeleton

We introduce a new topology-preserving 3D thinning procedure for deriving the curve voxel skeleton from 3D binary digital images. Based on a rigorously defined classification procedure, the algorithm consists of sequential thinning iterations each characterized by six parallel directional sub-iterations followed by a set of sequential sub-iterations. The algorithm is shown to produce concise and geometrically accurate 3D curve skeletons. The thinning algorithm is also insensitive to object rotation and only moderately sensitive to noise. Although this thinning procedure is valid for curve skeleton extraction of general elongated objects, in this paper, we specifically discuss its application to the orientation modeling of trabecular biological tissues.     

A real-time object detecting and tracking system for outdoor night surveillance

Autonomous video surveillance and monitoring has a rich history. Many deployed systems are able to reliably track human motion in indoor and controlled outdoor environments. However, object detection and tracking at night remain very important problems for visual surveillance. The objects are often distant, small and their signatures have low contrast against the background. Traditional methods based on the analysis of the difference between successive frames and a background frame will do not work. In this paper, a novel real time object detection algorithm is proposed for night-time visual surveillance. The algorithm is based on contrast analysis. In the first stage, the contrast in local change over time is used to detect potential moving objects. Then motion prediction and spatial nearest neighbor data association are used to suppress false alarms. Experiments on real scenes show that the algorithm is effective for night-time object detection and tracking.     

运动目标检测算法在军用视频监控系统中的应

在充分分析了现有算法在复杂背景下所存在的不足的基础上,提出了一种适用于军用视频监控系统的基于视频序列象素时空相关性检测的运动目标检测方法.该方法首先用每一帧中目标的空间相关性检测出目标;再用序列图像中目标的时间相关性检验目标的真实性,从而最终确定是否有运动目标.实际应用表明,采用该算法的系统能很好地检测出运动目标,并具有较强的抗干扰能力.     

基于改进光流特征的运动目标跟踪

在城市智能视频监控中需要对运动目标进行实时跟踪,针对传统的运动目标检测中出现的跟踪目标易丢失、跟踪率低、实时性差等问题,提出一种基于改进光流特征的运动目标跟踪检测方法,对运动行人目标进行跟踪。该方法首先采用改进的Vibe运动背景建模法对视频中存在的运动行人进行检测,再将Shi-Tomasi角点检测与LK光流法进行结合,将角点检测结果融入到LK光流法中,并对检测到的角点进行运动光流特征提取,最后通过卡尔曼滤波对出现的行人进行预测跟踪,采用匈牙利最优匹配算法实现对运动目标的持续匹配以及对运动目标的跟踪。仿真结果表明,本文提出的方法能够对视频中出现的运动目标进行检测跟踪,具有较好的识别效果,且检测效率得到提高。      

特征融合与S-D概率矫正的RGB-D显著检测

目的 许多先前的显著目标检测工作都是集中在2D的图像上,并不能适用于RGB-D图像的显著性检测。本文同时提取颜色特征以及深度特征,提出了一种基于特征融合和S-D概率矫正的RGB-D显著性检测方法,使得颜色特征和深度特征相互补充。方法 首先,以RGB图像的4个边界为背景询问节点,使用特征融合的Manifold Ranking输出RGB图像的显著图;其次,依据RGB图像的显著图和深度特征计算S-D矫正概率;再次,计算深度图的显著图并依据S-D矫正概率对该显著图进行S-D概率矫正;最后,对矫正后的显著图提取前景询问节点再次使用特征融合的Manifold Ranking方法进行显著优化,得到最终的显著图。结果 利用本文RGB-D显著性检测方法对RGBD数据集上的1 000幅图像进行了显著性检测,并与6种不同的方法进行对比,本文方法的显著性检测结果更接近人工标定结果。Precision-Recall曲线（PR曲线）显示在相同召回率下本文方法的准确率较其中5种方法高,且处理单幅图像的时间为2.150 s,与其他算法相比也有一定优势。结论 本文方法能较准确地对RGB-D图像进行显著性检测。     

低光照下的无人机异物检测与定位

为解决无人机在低光照环境下的巡检过程中, 不能对场景中的异物进行识别与定位, 导致后续智能算法无法获得环境语义信息的问题. 本文提出一种将ORB-SLAM2算法与适用于低光照目标检测改进的YOLOv5模型进行信息融合的方法. 首先, 通过RGB-D相机自采集低光照数据集进行深度学习训练及融合算法验证. 然后, 结合关键帧信息、目标检测模块的输出结果以及相机的固有信息完成目标像素坐标提取. 最后, 通过关键帧信息和像素坐标完成目标物体相对世界坐标系的位置解算. 本文实现了低光照环境下目标物体较为准确的识别和目标物体在世界坐标系中分米级的定位, 为低光照环境下无人机智能巡检提供了一种有效的解决方案.