双目视觉辅助的激光惯导SLAM算法

激光同步定位与地图构建(simultaneous localization and mapping, SLAM)算法在位姿估计和构建环境地图时依赖环境结构特征信息,在结构特征缺乏的场景下,此类算法的位姿估计精度与鲁棒性将下降甚至运行失败.对此,结合惯性测量单元(inertial measurement unit, IMU)不受环境约束、相机依赖视觉纹理的特点,提出一种双目视觉辅助的激光惯导SLAM算法,以解决纯激光SLAM算法在环境结构特征缺乏时的退化问题.即采用双目视觉惯导里程计算法为激光扫描匹配模块提供视觉先验位姿,并进一步兼顾视觉约束与激光结构特征约束进行联合位姿估计.此外,提出一种互补滤波算法与因子图优化求解的组合策略,完成激光里程计参考系与惯性参考系对准,并基于因子图将激光位姿与IMU数据融合以约束IMU偏置,在视觉里程计失效的情况下为激光扫描匹配提供候补的相对位姿预测.为进一步提高全局轨迹估计精度,提出基于迭代最近点匹配算法(iterative closest point, ICP)与基于图像特征匹配算法融合的混合闭环检测策略,利用6自由度位姿图优化方法显著降低里程计漂移误...     

A robust loop-closure method for visual SLAM in unstructured seafloor environments

This paper addresses the problem of visual simultaneous localization and mapping (SLAM) in an unstructured seabed environment that can be applied to an unmanned underwater vehicle equipped with a single monocular camera as the main measurement sensor. Monocular vision is regarded as an efficient sensing option in the context of SLAM, however it poses a variety of challenges when the relative motion is determined by matching a pair of images in the case of in-water visual SLAM. Among the various challenges, this research focuses on the problem of loop-closure which is one of the most important issues in SLAM. This study proposes a robust loop-closure algorithm in order to improve operational performance in terms of both navigation and mapping by efficiently reconstructing image matching constraints. To demonstrate and evaluate the effectiveness of the proposed loop-closure method, experimental datasets obtained in underwater environments are used, and the validity of the algorithm is confirmed by a series of comparative results.     

Matching by linear programming and successive convexification

We present a novel convex programming scheme to solve matching problems, focusing on the challenging problem of matching in a large search range and with cluttered background. Matching is formulated as metric labeling with L₁ regularization terms, for which we propose a novel linear programming relaxation method and an efficient successive convexification implementation. The unique feature of the proposed relaxation scheme is that a much smaller set of basis labels is used to represent the original label space. This greatly reduces the size of the searching space. A successive convexification scheme solves the labeling problem in a coarse to fine manner. Importantly, the original cost function is reconvexified at each stage, in the new focus region only, and the focus region is updated so as to refine the searching result. This makes the method well-suited for large label set matching. Experiments demonstrate successful applications of the proposed matching scheme in object detection, motion estimation, and tracking     

PIRF-Nav 2.0: Fast and online incremental appearance-based loop-closure detection in an indoor environment

This paper presents a fast and online incremental solution for an appearance-based loop-closure detection problem in a dynamic indoor environment. Closing the loop in a dynamic environment has been an important topic in robotics for decades. Recently, PIRF-Nav has been reported as being successful in achieving high recall rate at precision 1. However, PIRF-Nav has three main disadvantages: (i) the computational expense of PIRF-Nav is beyond real-time, (ii) it utilizes a large amount of memory in the redundant process of keeping signatures of places, and (iii) it is ill-suited to an indoor environment. These factors hinder the use of PIRF-Nav in a general environment for long-term, high-speed mobile robotic applications. Therefore, this paper proposes two techniques: (i) new modified PIRF extraction that makes the system more suitable for an indoor environment and (ii) new dictionary management that can eliminate redundant searching and conserve memory consumption. The results show that our proposed method can complete tasks up to 12 times faster than PIRF-Nav with only a slight percentage decline in recall. In addition, we collected additional data from a university canteen crowded during lunch time. Even in this crowded indoor environment, our proposed method has better real-time processing performance compared with other methods.     

室内环境下基于平面与线段特征的RGB-D视觉里程计

针对室内环境的结构特点,提出一种使用平面与线段特征的RGB-D视觉里程计算法．首先根据RGB-D扫描点的法向量对3D点云进行聚类,并使用随机抽样一致（RANSAC）算法对每簇3D点集进行平面拟合,抽取出环境中的平面特征;随后利用边缘点检测算法分割出环境中的边缘点集,并提取出环境中的线段特征;然后提出一种基于平面与线段几何约束的特征匹配算法,完成特征之间的匹配．在平面与线段特征匹配结果能提供充足的位姿约束的条件下,利用特征之间的匹配关系直接求解RGB-D相机的位姿;若不能,则利用匹配线段的端点以及线段点集来实现RGB-D相机位姿的估计．在TUM公开数据集中的实验证明了选择平面与线段作为环境特征可以提升视觉里程计估计和环境建图的精度．特别是在fr3/cabinet数据集中,本文算法的旋转、平移的均方根误差分别为2.046°/s、0.034m/s,要显著优于其他经典的视觉里程计算法．最终将本文系统应用到实际的移动机器人室内建图中,系统可以建立准确的环境地图,且系统运行速度可以达到3帧/s,满足实时处理的要求．     

基于快速增量式视觉感知的类脑SLAM

传统的RatSLAM算法中视觉处理受环境、光照的影响大,进而导致建图精度及稳定性下降。因此,提出了一种快速增量式视觉处理方法克服原RatSLAM系统中的视觉处理的缺陷。以一个改进型的二叉搜索树为检索算法,通过动态岛屿机制对图像进行分组,最终通过序列匹配的形式实现环境识别,达到了在线、准确、快速识别环境的目的。实验结果表明,所提算法的位置识别准确率高于99%,召回率高于80%,平均处理时间低于50ms。本系统的闭环性能、时间性能及建图稳定性均显著优于现有方案,进一步证明了基于快速增量式视觉处理方法的鲁棒性、高效性。     

结构化环境下基于结构单元软编码的3维激光雷达点云描述子

在同时定位与地图构建（SLAM）系统中,基于3维激光雷达点云数据的闭环检测由于描述子计算困难而极具挑战．为此,本文提出一种结构化环境下可用于闭环检测的基于结构单元软编码的新型3维激光雷达点云描述子．针对3维激光雷达点云的稀疏性和独立性导致的3维空间线段提取困难的问题,首先通过几何滤波的方法提取3维空间中垂直于地面的线段,用于保留3维空间的结构信息;然后,基于线段的空间几何关系构建结构单元集合,并通过软编码技术计算特征向量,作为3维激光雷达点云的描述子;最后,通过两帧点云描述子的匹配实现闭环检测．在KITTI公开数据集和自采数据集上的对比实验,验证了本文方法在时效性和鲁棒性等方面均优于主流的3维激光闭环检测方法．     

一种基于迭代的关系模型到本体模型的模式匹配方法

语义网的飞速发展，使得各领域出现了以本体这种形式来表达的知识模型.但在实际的语义网应用中，常常面临本体实例匮乏的问题.将现有关系型数据源中的数据转化为本体实例是一种有效的解决办法，这需要利用关系模型到本体模型的模式匹配技术来建立数据源和本体之间的映射关系.除此之外，关系模型到本体模型的模式匹配还被广泛用于数据集成、数据语义标注、基于本体的数据访问等领域中.现有的研究工作往往会综合使用多种模式匹配算法，计算异构数据模式中元素对的综合相似度，辅助人工建立数据源到本体的映射关系.现有的工作针对单一模式匹配算法准确率不高的问题，试图通过综合多种模式匹配算法的结果来进行调和.然而，这种方法当多种匹配算法同时出现不准时，难以得出更加准确的最终匹配结果.对单一模式匹配算法匹配不准的成因进行深入的分析，认为数据源的本地化特征是导致这一现象的重要因素，并提出了一种迭代优化的模式匹配方案.该方案利用在模式匹配过程中已经得到匹配的元素对，对单一模式匹配算法进行优化，经过优化后的算法能够更好地兼容数据源的本地化特征，从而显著提升准确率.在"餐饮信息管理"领域的一个实际案例上开展实验，模式匹配效果显著高于传统方法，其中，F值超过传统方法50.1%.     

Multi-scale feature extraction and adaptive matching for copy-move forgery detection

A copy-move forgery detection scheme by using multi-scale feature extraction and adaptive matching is proposed in this paper. First, the host image is segmented into the non-overlapping patches of irregular shape in different scales. Then, Scale Invariant Feature Transform is applied to extract feature points from all patches, to generate the multi-scale features. An Adaptive Patch Matching algorithm is subsequently proposed for finding the matching that indicate the suspicious forged regions in each scale. Finally, the suspicious regions in all scales are merged to generate the detected forgery regions in the proposed Matched Keypoints Merging algorithm. Experimental results show that the proposed scheme performs much better than the existing state-of-the-art copy-move forgery detection algorithms, even under various challenging conditions, including the geometric transforms, such as scaling and rotation, and the common signal processing, such as JPEG compression and noise addition; in addition, the special cases such as the multiple copies and the down-sampling are also evaluated, the results indicate the very good performance of the proposed scheme.     

一种基于半直接视觉里程计的RGB-D SLAM算法

提出了一种新的基于半直接视觉里程计的RGB-D SLAM（同步定位与地图创建）算法,同时利用直接法和传统特征点法的优势,结合鲁棒的后端优化和闭环检测,有效提高了算法在复杂环境中的定位和建图精度．在定位阶段,采用直接法估计相机的初始位姿,然后通过特征点匹配和最小化重投影误差进一步优化位姿,通过筛选地图点并优化位姿输出策略,使算法能够处理稀疏纹理、光照变化、移动物体等难题．算法具有全局重定位的能力．在后端优化阶段,提出了一种新的关键帧选取策略,同时保留直接法选取的局部关键帧和特征点法选取的全局关键帧,并行地维护2种关键帧,分别在滑动窗口和特征地图中对它们进行优化．算法通过对全局关键帧进行闭环检测和优化,提高SLAM的全局一致性．基于标准数据集和真实场景的实验结果表明,算法的性能在许多实际场景中优于主流的RGB-D SLAM算法,对纹理稀疏和有移动物体干扰的环境的鲁棒性较强．     

基于SSD算法的轻量化仪器表盘检测算法

在使用传统的图像识别算法对仪器表盘中的数字进行识别时,存在着流程繁琐,处理时间较长和检测效果不佳等问题。针对上述不足,提出了一种基于深度学习的轻量化仪器表盘检测算法,该算法以单发多尺度检测算法为基础,使用深度可分离卷积代替标准卷积设计特征提取网络,以提升特征表达能力和轻量化性能;同时提出了一种基于真实框分布构建锚框的流程,设计了能量化表达锚框匹配程度的指标——匹配率,促进构建匹配度更高且锚框数量更少的锚框方案。实验结果表明,所提算法具有较少的模型参数量和计算量,具有较高的检测精度,并且可在CPU环境下满足实时检测需求。     

一种广泛适用的激光雷达数据特征提取方法

提出了一种新颖的、无需先验知识的、广泛适用于各种环境的激光雷达数据特征提取方法来解决同步 定位与地图创建（SLAM）中的特征提取问题．这种方法采用经典的图像特征提取方法——Harris 角点探测器,具 体来说,是多尺度Kanade-Tomasi 角点探测器,来提取特征．这种方法可以从各种尺度的测量数据中提取稳定、精 确的特征点,并同时可以得到特征点描述器和不确定性信息．文章将这种方法应用在了软件仿真环境及经典数据集 上,包括：2 维的维多利亚公园数据集、英特尔研究中心数据集（Intel Research Center dataset）以及3 维的麻省理工 学院美国国防部高级研究计划局城市竞赛数据集（MIT DARPA Urban Challenge dataset）．实验结果表明这种方法可 以从各种环境中提取高精度、高重复性的稳定特征．     

基于相位一致特征的CBERS-02B遥感图像自动配准

以CBERS-02B卫星HR高分辨率和CCD多光谱遥感图像数据为基础,针对基于特征的自动配准方法中的特征检测与特征匹配两个关键步骤,通过引入性能优良的相位一致特征检测方法和特征相似与空间关系相结合的特征匹配策略,实现了一种基于相位一致特征的遥感图像高精度自动配准方法。实验结果表明,该方法对遥感图像亮度和对比度具有不变性,能稳定可靠提取HR高分辨率和CCD多光谱遥感图像显著的点特征,精确匹配相位一致特征点,实现了CBERS-02B卫星不同谱段,不同传感器和不同时相遥感图像间高精度自动配准,所进行实验的自动配准精度均到达了优于0.3 像元的系统配准精度。因此,该自动配准方法适合应用于有高配准精度要求的遥感图像间自动配准。     

Multiple target localization in wireless visual sensor networks

Target localization is an important service in wireless visual sensor networks (WVSN). Although the problem of single target localization has been intensively studied, few consider the problem of multiple target localization without prior target information in WVSN. In this paper, we first investigate the architecture of WVSN where data transmission is reduced to only target positions. Since target matching is a key issue in the multiple target localization, we propose a statistical method to match corresponding targets to located targets in world coordinates. In addition, we also consider scenarios where occlusion or limited field of view (FOV) occurs. The proposed method utilizes target images to the greatest extent. Our experimental results show that the proposed method obtains a more accurate result in targets localization compared with the camera discard scheme, and saves significant amounts of energy compared with other feature matching schemes.     

Feature mining and pattern classification for steganalysis of LSB matching steganography in grayscale images

In this paper, we present a scheme based on feature mining and pattern classification to detect LSB matching steganography in grayscale images, which is a very challenging problem in steganalysis. Five types of features are proposed. In comparison with other well-known feature sets, the set of proposed features performs the best. We compare different learning classifiers and deal with the issue of feature selection that is rarely mentioned in steganalysis. In our experiments, the combination of a dynamic evolving neural fuzzy inference system (DENFIS) with a feature selection of support vector machine recursive feature elimination (SVMRFE) achieves the best detection performance. Results also show that image complexity is an important reference to evaluation of steganalysis performance.     

针对重复模式图像的成对特征点匹配

目的 含有重复模式的图像会对局部特征描述符产生歧义,因此基于局部特征的匹配算法在此类图像的匹配过程中极易产生误匹配.同时,通过研究现有的引入全局特征描述符的匹配算法,发现全局特征同样依赖于计算局部信息所得到的特征点主方向,所以此类方法在含有重复模式的图像中也不容易得到令人满意的匹配效果.为了解决这一问题,提出一种基于成对特征点的图像匹配算法.方法 该方法利用成对特征点的方向向量作为特征点对的主方向,为特征描述提供了正确的方向信息,同时引入DAISY描述符与改进后的全局上下文(globalcontext)特征描述符,提高了匹配能力.结果 分别在模拟图像与实际图像上面进行了对比匹配实验,本文算法平均的匹配正确率能达到88％以上,比其他经典的匹配算法提高了26％以上.结论 实验结果表明,本文算法克服了现有算法在特征描述与主方向分配上的缺陷,进一步提升了匹配正确率,能够有效地解决重复模式图像的匹配问题.     

大数据下数据库引文索引匹配误差检测仿真

为了改善当前数据库引文索引匹配误差检测过程中受语境和歧义词影响,存在检测准确率不高、检测召回率较低、检测耗时较长的缺点,提出了基于APMI的大数据下数据库引文索引匹配误差检测方法。方法采用基于相对词频语境计算模型对数据库引文索引词语做分词消岐处理;采用TF-IDF方法提取数据库引用文献和被引用文献的特征词,并计算每个提取的特征词在语料库中的上下文概率分布,通过对比两个特征词在数据库中的上下文概率分布情况,将消岐后特征词之间的相对熵作为两篇参考文献中特征词之间的语义相似度;并以两篇参考文献中特征词之间的语义相似度平均值作为特征词和参考文献之间的匹配度;结合TF-IF方法计算大数据下数据库参考文献之间的匹配度,完成引文索引匹配误差检测。仿真测试结果显示,所提方法克服了当前方法受语境和歧义词影响的问题,具有较高的检测准确率、检测召回率,极大地缩短了检测耗时。     

Affine invariant watermarking algorithm using feature matching

Geometric attacks are among the most challenging problems in present day watermarking. Such attacks are very simple to implement yet they can defeat most of the existing watermarking algorithms without causing serious perceptual image distortion. In this research, a novel method to estimate the geometric manipulation is reported. The point pattern matching measure is computed for estimating geometric manipulation. The reference point patterns (i.e., a triple) are computed from feature ellipse of the original image. The proposed scheme does not require the original image because reference triple information of the original image has been contained in the secret key. Novel algorithm has been proved its robustness to geometric attacks through experiments results.     

基于视频指纹的快速视频拷贝检测方法

为满足视频拷贝检测的实时性要求,提出一种基于视频指纹的快速视频拷贝检测方法。基于DC图像序列提取改进的顺序度量特征和改进的运动特征,相结合生成视频指纹,通过视频指纹的相似性匹配对视频进行拷贝检测。实验结果证明,该方法能在保持准确性的同时快速地进行视频拷贝检测。