Evaluation of 3D interest point detection techniques via human-generated ground truth

In this paper, we present an evaluation strategy based on human-generated ground truth to measure the performance of 3D interest point detection techniques. We provide quantitative evaluation measures that relate automatically detected interest points to human-marked points, which were collected through a web-based application. We give visual demonstrations and a discussion on the results of the subjective experiments. We use a voting-based method to construct ground truth for 3D models and propose three evaluation measures, namely False Positive and False Negative Errors, and Weighted Miss Error to compare interest point detection algorithms.     

Illumination robust interest point detection

Most interest point detection algorithms are highly sensitive to illumination variations. This paper presents a method to find interest points robustly even under large non-uniform photometric changes. The method, which we call illumination robust feature extraction transform (IRFET), determines salient interest points in an image by calculating and analyzing contrast signatures. A contrast signature shows the response of an interest point detector with respect to a set of contrast stretching functions. The IRFET is generic and can be used with most interest point detectors. In this paper, we demonstrate that the IRFET improves the repeatability rate of the Harris corner detector significantly (by around 25% on average in the experiments).     

改进的基于冗余点过滤的3D目标检测方法

VoxelNet网络模型是第一个基于点云的端对端目标检测网络，只利用点云数据来生成高精度的3D目标检测框，具有十分良好的效果。但是，VoxelNet使用完整场景的点云数据作为输入，导致耗费了更多的计算资源在背景点云数据上，而且只包含几何信息的点云对目标的识别粒度较低，在较复杂的场景中容易出现误检测和漏检测。针对这些问题对VoxelNet进行了改进，在VoxelNet模型中加入视锥体候选区。首先，通过RGB前视图对感兴趣目标进行定位；然后，将目标2D位置升维至空间视锥体，在点云中提取目标视锥体候选区，过滤冗余点云，仅对视锥体候选区中的点云数据进行计算来得到检测结果。改进后的算法与VoxelNet相比，降低了点云计算量，避免了对背景点云数据的计算，提升了有效运算率，同时，避免了过多背景点的干扰，降低了误检测和漏检测率。KITTI数据集上的实验结果表明，改进后的算法在简单、中等、困难三种模式下的3D平均精度分别为67.92%、59.98%、53.95%，优于VoxelNet模型。     

改进的基于冗余点过滤的3D目标检测方法

VoxelNet网络模型是第一个基于点云的端对端目标检测网络,只利用点云数据来生成高精度的3D目标检测框,具有十分良好的效果。但是,VoxelNet使用完整场景的点云数据作为输入,导致耗费了更多的计算资源在背景点云数据上,而且只包含几何信息的点云对目标的识别粒度较低,在较复杂的场景中容易出现误检测和漏检测。针对这些问题对VoxelNet进行了改进,在VoxelNet模型中加入视锥体候选区。首先,通过RGB前视图对感兴趣目标进行定位;然后,将目标2D位置升维至空间视锥体,在点云中提取目标视锥体候选区,过滤冗余点云,仅对视锥体候选区中的点云数据进行计算来得到检测结果。改进后的算法与VoxelNet相比,降低了点云计算量,避免了对背景点云数据的计算,提升了有效运算率,同时,避免了过多背景点的干扰,降低了误检测和漏检测率。KITTI数据集上的实验结果表明,改进后的算法在简单、中等、困难三种模式下的3D平均精度分别为67.92%、59.98%、53.95%,优于VoxelNet模型。     

基于三维点云的阶梯目标检测与参数估计

阶梯目标检测与阶梯三维模型构建对移动机器人自主导航和运动规划具有重要意义.针对实际应用中阶梯目标结构的多样性以及点云分布的不确定性等特点,提出一种基于阶梯拓扑模型和模糊集理论的自适应阶梯目标检测与参数估计方法.利用阶梯剖面模型的拓扑关系与直方图算法,可有效提高阶梯边缘位置估计的精度及鲁棒性.采用同级线段提取与跨级线段接合策略,可实现对候选阶梯边缘线集合的准确构建.在此基础上,通过模糊变换和自适应模糊推理估计各级候选阶梯边缘线之间的级联概率,并采用模拟退火算法搜索全局最优的候选阶梯边缘线组合,从而实现对阶梯三维模型参数的有效估计.实验结果及数据分析验证了所提方法的有效性和实用性.     

基于图像与点云的三维障碍物检测

为提高平截头点云网络在三维障碍物检测中的精度,基于平截头点云网络的结构提出一种扩张平截头点云的检测方法。采用图像和点云数据,使用二维目标检测网络Yolov3,检测障碍物的二维包围框;扩张包围框的大小,在点云数据中提取出障碍物对应的点云;通过改进的Pointnet网络对该点云计算,得到障碍物的三维信息。在原模型基础上,加入扩张包围框,提高点云数据提取的完整性。通过KITTI数据集的验证和测试,实验结果表明,通过扩张二维包围框可以有效提高检测网络的性能。     

融合稀疏点云补全的3D目标检测算法

基于雷达点云的3D目标检测方法有效地解决了RGB图像的2D目标检测易受光照、天气等因素影响的问题.但由于雷达的分辨率以及扫描距离等问题,激光雷达采集到的点云往往是稀疏的,这将会影响3D目标检测精度.针对这个问题,提出一种融合稀疏点云补全的目标检测算法,采用编码、解码机制构建点云补全网络,由输入的部分稀疏点云生成完整的密...     

杂乱背景和摄像机移动下的时空兴趣点检测

提出了一种杂乱背景和摄像机移动场景下的时空兴趣点检测方法.结合空间域上的平滑和时间域上的平滑:在空间域上利用非线性各向异性扩散进行平滑,从而有效平滑掉噪声和杂乱背景中的细小结构,同时保持了人体动作的边缘信息;在时间域上利用了一个与梯度成正比的抑制函数对摄像机晃动/移动所导致的时间关系上的不确定性进行建模.本文方法有效地解决了复杂场景下时空兴趣点检测面临的杂乱背景和摄像机移动问题.     

伪激光点云增强的道路场景三维目标检测

目的 针对激光雷达点云稀疏性导致小目标检测精度下降的问题,提出一种伪激光点云增强技术,利用图像与点云融合,对稀疏的小目标几何信息进行补充,提升道路场景下三维目标检测性能。方法 首先,使用深度估计网络获取双目图像的深度图,利用激光点云对深度图进行深度校正,减少深度估计误差;其次,采用语义分割的方法获取图像的前景区域,仅将前景区域对应的深度图映射到三维空间中生成伪激光点云,提升伪激光点云中前景点的数量占比;最后,根据不同的观测距离对伪激光点云进行不同线数的下采样,并与原始激光点云进行融合作为最终的输入点云数据。结果 在KITTI （Karlsruhe Institute of Technology and Toyota Technological Institute at Chicago）数据集上的实验结果表明,该方法能够提升多个最新网络框架的小目标检测精度,以典型网络SECOND（sparselyembedded convolutional detection）、MVX-Net （multimodal voxelnet for 3D object detection）、Voxel-RCNN为例,在困难等级下,三维目标检测精度分别获得8.65%、7.32%和6.29%的大幅提升。结论 该方法适用于所有以点云为输入的目标检测网络,并显著提升了多个目标检测网络在道路场景下的小目标检测性能。该方法具备有效性与通用性。     

Violence detection in videos using interest frame extraction and 3D convolutional neural network

Multimedia Tools and Applications - With the rapid development of detecting violent behaviors in surveillance cameras, requests on systems that automatically recognize violent events are expanded....     

基于深度学习的点云三维目标检测方法综述

点云作为一种三维环境数据因其具有较高的精度一直被广泛关注并应用于多种场景任务之中。近年来,深度学习进入点云领域,让点云数据处理得到快速发展。针对基于深度学习的点云三维目标检测任务,首先分析了点云数据的特性并列举了日常任务中常用的点云数据集,随后通过单模态的三维目标检测与多模态的三维目标检测两个方向进行分类阐述,并通过单模态与多模态方法在数据集上的表现作比对。最后对当前点云三维目标检测研究的发展趋势进行展望与总结。     

Research on 3D medial axis transform via the saddle point programming method

The present paper investigates the 3D medial axis transform of objects bounded by freeform surfaces via the saddle point programming method, a mathematical programming approach used to identify the saddle points of a function. After exploring the local geometry and saddle point property of 3D medial axis transform, the mathematical programming method is employed to construct the saddle point programming models. Based on the optimality conditions that the optimal solutions should satisfy, a generic algorithm for computing various medial axis points is developed. In order to identify the junction points and localize the problem, the boundary and the skeletal curves are divided into skeletal segments, and it is proved to be efficient and accurate by numerical examples.     

Salient object detection via saliency bias and diffusion

Salient object detection aims to identify both spatial locations and scales of the salient object in an image. However, previous saliency detection methods generally fail in detecting the whole objects, especially when the salient objects are actually composed of heterogeneous parts. In this work, we propose a saliency bias and diffusion method to effectively detect the complete spatial support of salient objects. We first introduce a novel saliency-aware feature to bias the objectness detection for saliency detection on a given image and incorporate the saliency clues explicitly in refining the saliency map. Then, we propose a saliency diffusion method to fuse the saliency confidences of different parts from the same object for discovering the whole salient object, which uses the learned visual similarities among object regions to propagate the saliency values across them. Benefiting from such bias and diffusion strategy, the performance of salient object detection is significantly improved, as shown in the comprehensive experimental evaluations on four benchmark data sets, including MSRA-1000, SOD, SED, and THUS-10000.     

3D pedestrian tracking and frontal face image capture based on head point detection

Multimedia Tools and Applications - This paper proposes a method to track pedestrians in crowded scenes and capture the close-up frontal face images of a person of interest (POI) for recognition....     

An object detection algorithm combining semantic and geometric information of the 3D point cloud

Accurately detect vehicles or pedestrians from 3D point clouds (3D object detection) is a fast developing research topic in autonomous driving and other domains. The fundamental component for feature extraction in 3D object detection is Set Abstraction (SA), which can downsample points while aggregating points to extract features. However, the current SA ignores the geometric and semantic properties of point clouds and may miss to detect remote small objects. In this paper, FocusSA is proposed, which consists two modules for enhancing useful feature extraction in the SA layer to improve 3D object detection accuracy. At first, Focused FPS (FocFPS) is proposed to evaluate the foreground and boundary scores of the points and reweighs the Furthest Point Sampling (FPS) using the evaluated scores to retain more contextual points in downsampling. Then a Geometry-aware Feature Extraction (GeoFE) module is proposed to add geometric information to enrich the awareness of geometric structure in feature aggregation. To evaluate the performances of the proposed methods, we conduct extensive experiments on three difficulty levels of Car class in KITTI dataset. The experimental results show that on “moderate” instances, our results outperform the state-of-the-art method by 1.08%. Moreover, FocusSA is easy to be plugged in popular architectures.     

Robust registration of 2D and 3D point sets

This paper introduces a new method of registering point sets. The registration error is directly minimized using general-purpose non-linear optimization (the Levenberg–Marquardt algorithm). The surprising conclusion of the paper is that this technique is comparable in speed to the special-purpose Iterated Closest Point algorithm, which is most commonly used for this task. Because the routine directly minimizes an energy function, it is easy to extend it to incorporate robust estimation via a Huber kernel, yielding a basin of convergence that is many times wider than existing techniques. Finally, we introduce a data structure for the minimization based on the chamfer distance transform, which yields an algorithm that is both faster and more robust than previously described methods.     

E^2-PV-RCNN: improving 3D object detection via enhancing keypoint features

Multimedia Tools and Applications - 3D object detection plays a vital role and exerts a growing important effect in many applications, such as 3D scene understanding and autonomous driving. In this...