A robust omnidirectional vision sensor for soccer robots

How to make robot vision work robustly under varying lighting conditions and without the constraint of the current color-coded environment are two of the most challenging issues in the RoboCup community. In this paper, we present a robust omnidirectional vision sensor to deal with these issues for the RoboCup Middle Size League soccer robots, in which two novel algorithms are applied. The first one is a camera parameters auto-adjusting algorithm based on image entropy. The relationship between image entropy and camera parameters is verified by experiments, and camera parameters are optimized by maximizing image entropy to adapt the output of the omnidirectional vision to the varying illumination. The second one is a ball recognition method based on the omnidirectional vision without color classification. The conclusion is derived that the ball on the field can be imaged to be an ellipse approximately in our omnidirectional vision, and the arbitrary FIFA ball can be recognized by detecting the ellipse imaged by the ball. The experimental results show that a robust omnidirectional vision sensor can be realized by using the two algorithms mentioned above.     

利用稀疏化生物视觉特征的多类多视角目标检测方法

鉴于生物视觉特征在目标分类中的良好性能,采用了一种基于稀疏化生物视觉特征的多类多视角目标检测方法。首先利用特征稀疏化方法对生物视觉的标准模型进行改进,有效提高了目标的可分性,然后利用滑动窗口的方法构建基于稀疏化生物视觉特征的目标检测器,采用局部邻域抑制算法完成特定目标的检测任务,最后通过构建场景中待检测目标的词典,对每类目标分别设计滑动检测器以完成多类多视角目标的检测任务,实验结果表明,该方法具有很好的检测性能。     

Wide-angle and long-range real time pose estimation: A comparison between monocular and stereo vision systems

In this work, a comparison of the performances of a stereo and a monocular vision system for the 3D pose estimation of a planar target in very challenging conditions is presented. In particular, the systems have been designed to detect in real time a target moving with a maximum speed of 1 m/s, in a range of distances from 0.5 to 4 m from the cameras, with an accuracy of less than 1 cm (referred to the estimation of the real world coordinates) and with a field of view of 80 deg. A theoretical evaluation and experimental results to assess the performance of the proposed systems are presented. Our analysis demonstrates the good accuracy in terms of target position estimation of the presented approaches not only for close range applications, but also for mid-to-long range ones.     

多传感器线结构光视觉测量系统全局校准

针对现有多传感器线结构光视觉测量系统全局校准过程中通常需要三维测量设备、校准过程复杂等问题,提出了一种基于一维靶标的多传感器线结构光视觉测量系统全局校准方法。此方法以一维靶标为中介,首先由每个线结构光视觉传感器得到光条与一维靶标相交点在各自测量坐标系下的三维坐标,然后由交比不变性得到相交点在一维靶标坐标系下的坐标,进而...     

火炮身管结构光检测系统标定方法研究

火炮身管结构光检测系统工作环境视场小、光线弱,导致传统的标定方法存在较大的局限性。针对这一问题,应用结构光成像原理,提出了一种现场标定方法。以155口径火炮为例,设计了内表面包含多种结构信息的标定筒,利用该方法对标定筒内表面结构光图像进行处理,得到了标定筒内凹槽深度与结构光条纹偏移像素距离之间的关系,进而标定了检测系统结构参数。利用标定完成的检测系统对火炮身管膛线的磨损量进行检测,结果表明,检测绝对误差不超过0.01 mm,符合火炮现场检测的要求。     

基于机器视觉的老年人摔倒检测系统

红外图像中的人体摔倒检测不受环境光照射的影响,在智能安防领域有着重要的研究意义和应用价值。现有的摔倒检测方法没有充分考虑人体关键点的位置变化规律,容易对类摔倒动作造成误检。针对这一问题,本文提出一种基于改进Alphapose的红外图像摔倒检测算法。该算法使用Yolo v5s目标检测网络,在提取人体目标框输入姿态估计网络的同时,对人体姿态进行直接分类,再结合人体骨架关键点的位置信息和姿态特征进行判断。通过实验证明,该算法在准确度和实时性方面都有良好的表现。     

一种基于机器视觉的铝管缺陷检测系统

针对铝管生产过程中对准确、可量化的自动缺陷检测系统的迫切需要,本文引入一种由图像采集、缺陷检测、缺陷处理等模块组成的铝管缺陷检测系统。平板探测器获取由X光高压电源产生,穿过铝管的X射线并把所形成的数字图像通过USB端口发送至检测服务器。检测服务器使用机器视觉算法检测图像中的缺陷。当服务器检测到缺陷时,会向PCI板上指定位输出信号,报警装置接到信号后报警提醒工作人员。实验表明该系统能够自动、准确的标记出铝管中存在的缺陷,达到了系统的设计目标。     

Interoperable vision component for object detection and 3D pose estimation for modularized robot control

Finding objects and tracking their poses are essential functions for service robots, in order to manipulate objects and interact with humans. We present novel algorithms for local feature matching for object detection, and 3D pose estimation. Our feature matching algorithm takes advantage of local geometric consistency for better performance, and the new 3D pose estimation algorithm solves the pose in a closed-form using homography, followed by a non-linear optimization step for stability. Advantages of our approach include better performance, minimal prior knowledge for the target pattern, and easy implementation and portability as a modularized software component. We have implemented our approach along with both CPU and GPU-based feature extraction, and built an interoperable component that can be used in any Robot Technology (RT)-based control system. Experiment shows that our approach produces very robust results for the estimated 3D pose, and maintain very low false positive rate. It is also fast enough to be used in on-line applications. We integrated our vision component in an autonomous robot system with a search-and-grasp task, and tested it with several objects that are found in ordinary domestic environment. We present the details of our approach, the design of our modular component design, and the results of the experiments in this paper.     

基于机器视觉的凸轮轴表面缺陷检测系统

传统的凸轮轴表面缺陷多为产后抽检,相比在线全检存在漏检与缺陷率上升等现象,且只能事后发现;而人工在线全检不但会使成本上升、也对人力资源提出了考验。为此实现自动实时在线全检就成为急需解决的课题。设计了基于机器视觉的凸轮轴表面缺陷在线自动检测系统。系统安装在凸轮轴生产流水线两侧,搭建特定光源,在凸轮轴移动、停止、旋转过程中通过高速相机对其表面进行图像捕获,并由工控机进行缺陷判定与定位。根据轴类表面缺陷的特征,设计了缺陷分割算法和缺陷区域标记算法,对凸轮轴表面的外伤、砂眼、研磨不良等典型缺陷进行分辨。算法可以准确提取目标缺陷区域,标记缺陷位置并统计缺陷特征对缺陷进行判定。该系统可在0.44 s每根轴的速度下,检测出凸轮轴表面直径大于1 mm 的缺陷,并通过人机交互界面显示缺陷所在位置。完全可以取代产后抽检及人工在线全检,同时还可以提高检测效率与检测精度。     

An efficient forgery detection algorithm for object removal by exemplar-based image inpainting

As a popular image manipulation technique, object removal can be achieved by image-inpainting without any noticeable traces, which poses huge challenges to passive image forensics. The existing detection approach utilizes full search for block matching, resulting in high computational complexity. This paper presents an efficient forgery detection algorithm for object removal by exemplar-based inpainting, which integrates central pixel mapping (CPM), greatest zero-connectivity component labeling (GZCL) and fragment splicing detection (FSD). CPM speeds up suspicious block search by efficiently matching those blocks with similar hash values and then finding the suspicious pairs. To improve the detection precision, GZCL is used to mark the tampered pixels in suspected block pairs. FSD is adopted to distinguish and locate tampered regions from its best-match regions. Experimental results show that the proposed algorithm can reduce up to 90% of the processing time and maintain a detection precision above 85% under different kinds of object-removed images.     

SCPNet: Self-constrained parallelism network for keypoint-based lightweight object detection

Keypoint-based object detection achieves better performance without positioning calculations and extensive prediction. However, they have heavy backbone, and high-resolution is restored using upsampling that obtain unreliable features. We propose a self-constrained parallelism keypoint-based lightweight object detection network (SCPNet), which speeds inference, drops parameters, widens receptive fields, and makes prediction accurate. Specifically, the parallel multi-scale fusion module (PMFM) with parallel shuffle blocks (PSB) adopts parallel structure to obtain reliable features and reduce depth, adopts repeated multi-scale fusion to avoid too many parallel branches. The self-constrained detection module (SCDM) has a two-branch structure, with one branch predicting corners, and employing entad offset to match high-quality corner pairs, and the other branch predicting center keypoints. The distances between the paired corners’ geometric centers and the center keypoints are used for self-constrained detection. On MS-COCO 2017 and PASCAL VOC, SCPNet’s results are competitive with the state-of-the-art lightweight object detection. https://github.com/mengdie-wang/SCPNet.git.     

红外目标检测网络改进半监督迁移学习方法

针对红外数据集规模小,标记样本少的特点,提出了一种红外目标检测网络的半监督迁移学习方法,主要用于提高目标检测网络在小样本红外数据集上的训练效率和泛化能力,提高深度学习模型在训练样本较少的红外目标检测等场景当中的适应性。文中首先阐述了在标注样本较少时无标注样本对提高模型泛化能力、抑制过拟合方面的作用。然后提出了红外目标检测网络的半监督迁移学习流程:在大量的RGB图像数据集中训练预训练模型,后使用少量的有标注红外图像和无标注红外图像对网络进行半监督学习调优。另外,文中提出了一种特征相似度加权的伪监督损失函数,使用同一批次样本的预测结果相互作为标注,以充分利用无标注图像内相似目标的特征分布信息;为降低半监督训练的计算量,在伪监督损失函数的计算中,各目标仅将其特征向量邻域范围内的预测目标作为伪标注。实验结果表明,文中方法所训练的目标检测网络的测试准确率高于监督迁移学习所获得的网络,其在Faster R-CNN上实现了1.1%的提升,而在YOLO-v3上实现了4.8%的显著提升,验证了所提出方法的有效性。     

条纹投影轮廓术系统模型与标定综述

条纹投影轮廓术能较好地兼顾系统灵活性与测量精度,是光学三维表面成像与测量的主流技术。利用条纹投影轮廓术进行三维成像,首先需要建立合适的系统模型,然后通过系统标定来确定描述模型的系统参数,最后利用标定的系统模型进行三维重建,获得物体的三维表面形貌。由此可见,系统标定与系统模型密不可分,对三维成像的性能有直接影响。根据相位-三维映射和双目立体视觉两类不同的工作原理,对条纹投影轮廓术的系统模型和系统标定方法进行了综述,并简要总结了评估系统精度的方法和依据。     

一种基于计算机视觉的低成本室内空间测量系统设计与实现

设计并实现一种基于计算机视觉的室内空间测量系统,通过3个低成本的USB摄像头对室内目标进行图像拍摄,实时测量出目标的空间精确位置。采用张氏平面标定法标定摄像头的内外参数,并针对室内环境特点设计了简单有效的图像处理方法,利用立体视觉定位原理计算出目标点的空间位置信息。实验结果表明：系统测量误差小于15 mm,数据更新率为15 Hz,均可满足室内机器人的定位需要。     

GCENet: Global contextual exploration network for RGB-D salient object detection

Representing contextual features at multiple scales is important for RGB-D SOD. Recently, due to advances in backbone convolutional neural networks (CNNs) revealing stronger multi-scale representation ability, many methods achieved comprising performance. However, most of them represent multi-scale features in a layer-wise manner, which ignores the fine-grained global contextual cues in a single layer. In this paper, we propose a novel global contextual exploration network (GCENet) to explore the performance gain of multi-scale contextual features in a fine-grained manner. Concretely, a cross-modal contextual feature module (CCFM) is proposed to represent the multi-scale contextual features at a single fine-grained level, which can enlarge the range of receptive fields for each network layer. Furthermore, we design a multi-scale feature decoder (MFD) that integrates fused features from CCFM in a top-down way. Extensive experiments on five benchmark datasets demonstrate that the proposed GCENet outperforms the other state-of-the-art (SOTA) RGB-D SOD methods.     

基于机器视觉的香烟小包装外观质量检测系统

讨论了基于机器视觉的香烟小包装外观质量在线检测系统总体方案的设计,并针对香烟小包装流水线的具体情况,分别对香烟小包的检测位置以及图像采集方案进行了探讨.文中采用了一种用于香烟包装质量检测的快速图像匹配算法,并利用开源计算机视觉库OpenCV进行算法实现,仿真实验结果表明,该图像匹配算法计算速度快、检测精度高,满足香烟包装质量检测的需要.     

一种基于机器视觉的视频交通流检测系统

简要分析了交通流检测技术的发展现状,结合当前智能交通系统的应用需求,利用连续三帧差分的运动估计方法来构建初始背景,并采用统计打分的策略实时地对背景进行更新;同时提出了一种简单而有效的阴影消除算法以提高交通流参数检测的准确度。另外,针对现有交通流检测系统无车辆跟踪这一环节,可能导致流量多计数的问题,本文提出同时利用车辆的位置信息、颜色信息和分形维信息对车辆进行匹配跟踪的策略。大量实验证明该检测算法能快速、有效地检测出各种交通流参数,为实现交通管理的自动化奠定基础。     

SRI-Net: Similarity retrieval-based inference network for light field salient object detection

The cutting-edge RGB saliency models are prone to fail for some complex scenes, while RGB-D saliency models are often affected by inaccurate depth maps. Fortunately, light field images can provide a sufficient spatial layout depiction of 3D scenes. Therefore, this paper focuses on salient object detection of light field images, where a Similarity Retrieval-based Inference Network (SRI-Net) is proposed. Due to various focus points, not all focal slices extracted from light field images are beneficial for salient object detection, thus, the key point of our model lies in that we attempt to select the most valuable focal slice, which can contribute more complementary information for the RGB image. Specifically, firstly, we design a focal slice retrieval module (FSRM) to choose an appropriate focal slice by measuring the foreground similarity between the focal slice and RGB image. Secondly, in order to combine the original RGB image and the selected focal slice, we design a U-shaped saliency inference module (SIM), where the two-stream encoder is used to extract multi-level features, and the decoder is employed to aggregate multi-level deep features. Extensive experiments are conducted on two widely used light field datasets, and the results firmly demonstrate the superiority and effectiveness of the proposed SRI-Net.