单目同时定位与建图中的地图恢复融合技术

目的 传统的单目视觉SLAM（simultaneous localization and mapping）跟踪失败后需要相机重新回到丢失的位置才能重定位并恢复建图,这极大限制了单目SLAM的应用场景。为解决这一问题,提出一种基于视觉惯性传感器融合的地图恢复融合算法。方法 当系统跟踪失败,仅由惯性传感器提供相机位姿,通过对系统重新初始化并结合惯性传感器提供的丢失部分的相机位姿将丢失前的地图融合到当前的地图中;为解决视觉跟踪丢失期间由惯性测量计算导致的相机位姿误差,提出了一种以关键帧之间的共视关系为依据的跳跃式的匹配搜索策略,快速获得匹配地图点,再通过非线性优化求解匹配点之间的运动估计,进行误差补偿,获得更加准确的相机位姿,并删减融合后重复的点云;最后建立前后两个地图中关键帧之间与地图点之间的联系,用于联合优化后续的跟踪建图过程中相机位姿和地图点位置。结果 利用Euroc数据集及其他数据进行地图精度和地图完整性实验,在精度方面,将本文算法得到的轨迹与ground truth和未丢失情况下得到的轨迹进行对比,结果表明,在SLAM系统跟踪失败的情况下,此方法能有效解决系统无法继续跟踪建图的问题,其精度可达厘米级别。在30 m²的室内环境中,仅有9 cm的误差,而在300 m²工厂环境中误差仅有7 cm。在完整性方面,在相机运动较剧烈的情况下,恢复地图的完整性优于ORB_SLAM的重定位算法,通过本文算法得到的地图关键帧数量比ORB_SLAM多30%。结论 本文提出的算法在单目视觉SLAM系统跟踪失败之后,仍然能够继续跟踪建图,不会丢失相机轨迹。此外,无需相机回到丢失之前的场景中,只需相机观察到部分丢失前场景,即可恢复融合所有地图。本文算法不仅保证了恢复地图的精度,还保证了建图的完整性。与传统的重定位方法相比,本文算法在系统建图较少时跟踪失败的情况下效果更好。     

Polynomial algorithms for subisomorphism of nD open combinatorial maps

Combinatorial maps describe the subdivision of objects in cells, and incidence and adjacency relations between cells, and they are widely used to model 2D and 3D images. However, there is no algorithm for comparing combinatorial maps, which is an important issue for image processing and analysis. In this paper, we address two basic comparison problems, i.e., map isomorphism, which involves deciding if two maps are equivalent, and submap isomorphism, which involves deciding if a copy of a pattern map may be found in a target map. We formally define these two problems for nD open combinatorial maps, we give polynomial time algorithms for solving them, and we illustrate their interest and feasibility for searching patterns in 2D and 3D images, as any child would aim to do when he searches Wally in Martin Handford’s books.     

Improving local forest volume estimates by fusion of multi-temporal forest type maps

An alternative method for estimating standing wood volume based on the fusion of multi-temporal forest type maps and single reduced associated ground-based inventories is proposed. With the integration of photo-interpreted forest map realizations from different years into a single “Fused Map” resulting in an improved local estimate of the forest type, the proposed method offers more accurate estimates than the approach traditionally used by the Quebec Ministry of Natural Resources, even with reduced ground-based inventory effort. A Fused Map is basically an adapted mean of local forest type that accounts for differences among the classification systems used for each map, temporal differences between maps, and subjectivity associated with photo-interpreted data.     

Multiple path exploration for graph matching

The graph matching problem is a hot topic in machine vision. Although a myriad of matching algorithms have been proposed during decades of investigation, it is still a challenging issue because of the combinatorial nature. As one of the outstanding graph matching algorithms, the graduated nonconvexity and concavity procedure follows the path following algorithm. The main drawback of this approach lies that there may exist singular points which violate the smoothness of the solution path and thus harm the accuracy of matching. Addressing this problem, we develop a novel algorithm to bypass this pitfall to improve the matching accuracy. We design an effective method of singular point discovering by checking the smoothness of the path and subsequently explore multiple smooth curves at detected points for better matching results. For evaluation, we make comparisons between our approach and several outstanding matching algorithms on three popular benchmarks, and the results reveal the advantage of our approach.     

Finding approximate palindromes in strings

We introduce a novel definition of approximate palindromes in strings, and provide an algorithm to find all maximal approximate palindromes in a string with up to k errors. Our definition is based on the usual edit operations of approximate pattern matching, and the algorithm we give, for a string of size n on a fixed alphabet, runs in O(k²n) time. We also discuss two implementation-related improvements to the algorithm, and demonstrate their efficacy in practice by means of both experiments and an average-case analysis.     

Multi-VMap: A Multi-Scale Model for Vector Maps

Multi-VMap is a compact framework from which plane graphs representing geographic maps at different levels of detail can be extracted. Its main feature is that the scale of the extracted map can be variable through its domain, while each entity maintains consistent combinatorial relations with the rest of entities represented in the map. The model is based on a set of operators, called updates, which modify the level of detail in a small portion of a map. The set of updates is partially ordered, and can therefore be represented as a Directed Acyclic Graph, which defines our multi-scale structure. An algorithm to extract a map at the required resolution is proposed, and a lower bound for the number of different maps which can be extracted from the model is given. The model supports map data processing operations (e.g., querying), as well as progressive and selective transmission of maps over a network.     

Attributed relational graph matching based on the nested assignment structure

In this paper, we propose a new ARG matching scheme based on the nested assignment structure to assess the similarity between two attributed relational graphs (ARGs). ARGs are represented by nodes and edges containing unary attributes and binary relations between nodes, respectively. The nested assignment structure consists of inner and outer steps. In the inner step, to form a distance matrix, combinatorial differences between every pair of nodes in two ARGs are computed by using an assignment algorithm. Then, in the outer step, a correspondence between nodes in the two ARGs is established by using an assignment algorithm based on the distance matrix. The proposed ARG matching scheme consists of three procedures as follows: first, in the initializing procedure, the nested assignment structure is performed to generate an initial correspondence between nodes in two ARGs. Next, the correspondence is refined by iteratively performing the updating procedure, which also utilizes the nested assignment structure, until the correspondence does not change. Finally, the verifying procedure can be performed in case that some nodes to be matched in the two ARGs are missing. From experimental results, the proposed ARG matching scheme shows superior matching performance and localizes target objects robustly and correctly even in severely noisy and occluded scenes.     

Shape modeling and matching in identifying 3D protein structures

In this paper, we describe a novel geometric approach in the process of recovering 3D protein structures from scalar volumes. The input to our method is a sequence of α-helices that make up a protein, and a low-resolution protein density volume where possible locations of α-helices have been detected. Our task is to identify the correspondence between the two sets of helices, which will shed light on how the protein folds in space. The central theme of our approach is to cast the correspondence problem as that of shape matching between the 3D volume and the 1D sequence. We model both shapes as attributed relational graphs, and formulate a constrained inexact graph matching problem. To compute the matching, we developed an optimal algorithm based on the A*-search with several choices of heuristic functions. As demonstrated in a suite of synthetic and authentic inputs, the shape-modeling approach is capable of identifying helix correspondences in noise-abundant volumes at high accuracy with minimal or no user intervention.     

Fast spatio-temporal stereo for intelligent transportation systems

This paper presents a fast approach for matching stereoscopic images acquired by stereo cameras mounted aboard a moving car. The proposed approach exploits the spatio-temporal consistency between consecutive frames in stereo sequences to improve matching results. This means that the matching process at current frame uses the matching results obtained at its preceding one. The preceding frame allows to compute an Initial Disparity Map for the current frame. The initial disparity map is used to derive disparity ranges for each scanline as well as what we call Matching Control Edge Points. Dynamic programming is performed for matching edge points in stereo pairs. The matching control edge points are used to drive the search for an optimal solution in the search plane. This is accomplished by dividing the dynamic programming search space into a number of subspaces depending on the number of the matching control edge points. The proposed approach has been tested both on virtual and real stereo images sequences demonstrating satisfactory performance.     

A Fully Compressed Algorithm for Computing the Edit Distance of Run-Length Encoded Strings

A recent trend in stringology explores the possibility of utilizing text compression to speed up similarity computation between strings. In this line of investigation, run-length encoding is one of the earliest studied compression schemes. Despite its simple coding nature, the only positive result before this work is the computation of the in-del distance (dual of longest common subsequence), which requires O(mnlogmn) time, where m and n denote the number of runs of the input strings. The worst-case time complexity of computing the edit distance between two run-length encoded strings still depends on the uncompressed string lengths. In this paper, we break the foundational gap by providing its first “fully compressed” algorithm whose running time depends solely on the compressed string lengths. Specifically, given two strings, compressed into m and n runs, m≤n, we present an O(mn ²)-time algorithm for computing the edit distance of the strings. Our approach also yields the first fully compressed solution to approximate matching of a pattern of m runs in a text of n runs in O(mn ²) time.     

Elevation moment of inertia: A new feature for Monte Carlo localization in outdoor environment with elevation map

The elevation map is one of the most popular maps for outdoor navigation. We propose the elevation moment of inertia (EMOI), which represents the distribution of elevation around a robot in an elevation map, for use in the matching of elevation maps. Using this feature, outdoor localization can be performed with an elevation map without external positioning systems. In this research, the Monte Carlo localization (MCL) method is used for outdoor localization, and the conventional method is based on range matching, which compares range sensor data with the range data predicted from an elevation map. Our proposed method is based on EMOI matching. The EMOI around a robot is compared with the EMOIs for all cells of the pregiven reference elevation map to find a robot pose with respect to the reference map. MCL based on EMOI matching is very fast, although its accuracy is slightly lower than that of conventional range matching. To deal with the disadvantage of EMOI matching, an adaptive switching scheme between EMOI matching and range matching was also proposed. Various outdoor experiments indicated that the proposed EMOI significantly reduced the convergence time of MCL. Therefore, the proposed feature is considered to be useful when an elevation map is used for outdoor localization. © 2010 Wiley Periodicals, Inc.     

Learning low-rank and discriminative dictionary for image classification

Dictionary learning plays a crucial role in sparse representation based image classification. In this paper, we propose a novel approach to learn a discriminative dictionary with low-rank regularization on the dictionary. Specifically, we apply Fisher discriminant function to the coding coefficients to make the dictionary more discerning, that is, a small ratio of the within-class scatter to between-class scatter. In practice, noisy information in the training samples will undermine the discriminative ability of the dictionary. Inspired by the recent advances in low-rank matrix recovery theory, we apply low-rank regularization on the dictionary to tackle this problem. The iterative projection method (IPM) and inexact augmented Lagrange multiplier (ALM) algorithm are adopted to solve our objective function. The proposed discriminative dictionary learning with low-rank regularization (D²L²R²) approach is evaluated on four face and digit image datasets in comparison with existing representative dictionary learning and classification algorithms. The experimental results demonstrate the superiority of our approach.     

Unified Compression-Based Acceleration of Edit-Distance Computation

The edit distance problem is a classical fundamental problem in computer science in general, and in combinatorial pattern matching in particular. The standard dynamic programming solution for this problem computes the edit-distance between a pair of strings of total length O(N) in O(N ²) time. To this date, this quadratic upper-bound has never been substantially improved for general strings. However, there are known techniques for breaking this bound in case the strings are known to compress well under a particular compression scheme. The basic idea is to first compress the strings, and then to compute the edit distance between the compressed strings. As it turns out, practically all known o(N ²) edit-distance algorithms work, in some sense, under the same paradigm described above. It is therefore natural to ask whether there is a single edit-distance algorithm that works for strings which are compressed under any compression scheme. A rephrasing of this question is to ask whether a single algorithm can exploit the compressibility properties of strings under any compression method, even if each string is compressed using a different compression. In this paper we set out to answer this question by using straight line programs. These provide a generic platform for representing many popular compression schemes including the LZ-family, Run-Length Encoding, Byte-Pair Encoding, and dictionary methods. For two strings of total length N having straight-line program representations of total size n, we present an algorithm running in O(nNlg(N/n)) time for computing the edit-distance of these two strings under any rational scoring function, and an O(n ^2/3 N ^4/3) time algorithm for arbitrary scoring functions. Our new result, while providing a speed up for compressible strings, does not surpass the quadratic time bound even in the worst case scenario.     

基于主动环形闭合约束的移动机器人分层同时定位和地图创建

基于Rao-Blackwellized 粒子滤波器提出了一种基于主动闭环策略的移动机器人分层同时定位和地图创建(simultaneous localization and mapping, SLAM)方法,基于信息熵的主动闭环策略同时考虑机器人位姿和地图的不确定性;局部几何特征地图之间的相对关系通过一致性算法估计,并通过环形闭合约束的最小化过程回溯修正.在仅有单目视觉和里程计的基础上,建立了鲁棒的感知模型;通过有效的尺度不变特征变换(scale invariant feature transform, SIFT)方法提取环境特征,基于KD-Tree的最近邻搜索算法实现特征匹配.实际实验表明该方法为实现SLAM提供了一种有效可靠的途径.     

Integrating fuzzy topological maps and fuzzy geometric maps for behavior‐based robots

In behavior‐based robots, planning is necessary to elaborate abstract plans that resolve complex navigational tasks. Usually maps of the environment are used to plan the robot motion and to resolve the navigational tasks. Two types of maps have been mainly used: metric and topological maps. Both types present advantages and weakness so that several integration approaches have been proposed in literature. However, in many approaches the integration is conducted to build a global representation model, and the planning and navigational techniques have not been fitted to profit from both kinds of information. We propose the integration of topological and metric models into a hybrid deliberative‐reactive architecture through a path planning algorithm based on A^* and a hierarchical map with two levels of abstraction. The hierarchical map contains the required information to take advantage of both kinds of modeling. On one hand, the topological model is based on a fuzzy perceptual model that allows the robot to classify the environment in distinguished places, and on the other hand, the metric map is built using regions of possibility with the shape of fuzzy segments, which are used later to build fuzzy grid‐based maps. The approach allows the robot to decide on the use of the most appropriate model to navigate the world depending on minimum‐cost and safety criteria. Experiments in simulation and in a real office‐like environment are shown for validating the proposed approach integrated into the navigational architecture. © 2002 Wiley Periodicals, Inc.     

Graph matching based on spectral embedding with missing value

This paper proposes an efficient algorithm for inexact graph matching based on spectral embedding with missing value. We commence by building an association graph model based on initial matching algorithm. Then, by dot product representation of graph with missing value, a new embedding method (co-embedding), where the correspondences between unmatched nodes are treated as missing data in an association graph, is presented. At last, a new graph matching algorithm which alternates between the co-embedding and point pattern matching is proposed. Convictive experimental results on both synthetic and real-world data demonstrate the effectiveness of the proposed graph matching algorithm.     

On the performance of self-organizing maps for the non-Euclidean Traveling Salesman Problem in the polygonal domain

In this paper, two state-of-the-art algorithms for the Traveling Salesman Problem (TSP) are examined in the multi-goal path planning problem motivated by inspection planning in the polygonal domain W. Both algorithms are based on the self-organizing map (SOM) for which an application in W is not typical. The first is Somhom’s algorithm, and the second is the Co-adaptive net. These algorithms are augmented by a simple approximation of the shortest path among obstacles in W. Moreover, the competitive and cooperative rules are modified by recent adaptation rules for the Euclidean TSP, and by proposed enhancements to improve the algorithms’ performance in the non-Euclidean TSP. Based on the modifications, two new variants of the algorithms are proposed that reduce the required computational time of their predecessors by an order of magnitude, therefore making SOM more competitive with combinatorial heuristics. The results show how SOM approaches can be used in the polygonal domain so they can provide additional features over the classical combinatorial approaches based on the complete visibility graph.     

Deblurring and Denoising of Maps between Shapes

Shape correspondence is an important and challenging problem in geometry processing. Generalized map representations, such as functional maps, have been recently suggested as an approach for handling difficult mapping problems, such as partial matching and matching shapes with high genus, within a generic framework. While this idea was shown to be useful in various scenarios, such maps only provide low frequency information on the correspondence. In many applications, such as texture transfer and shape interpolation, a high quality pointwise map that can transport high frequency data between the shapes is required. We name this problem map deblurring and propose a robust method, based on a smoothness assumption, for its solution. Our approach is suitable for non‐isometric shapes, is robust to mesh tessellation and accurately recovers vertex‐to‐point, or precise, maps. Using the same framework we can also handle map denoising, namely improvement of given pointwise maps from various sources. We demonstrate that our approach outperforms the state‐of‐the‐art for both deblurring and denoising of maps on benchmarks of non‐isometric shapes, and show an application to high quality intrinsic symmetry computation.     

Good edit similarity learning by loss minimization

Similarity functions are a fundamental component of many learning algorithms. When dealing with string or tree-structured data, measures based on the edit distance are widely used, and there exist a few methods for learning them from data. However, these methods offer no theoretical guarantee as to the generalization ability and discriminative power of the learned similarities. In this paper, we propose an approach to edit similarity learning based on loss minimization, called GESL. It is driven by the notion of (?,??,??)-goodness, a theory that bridges the gap between the properties of a similarity function and its performance in classification. Using the notion of uniform stability, we derive generalization guarantees that hold for a large class of loss functions. We also provide experimental results on two real-world datasets which show that edit similarities learned with GESL induce more accurate and sparser classifiers than other (standard or learned) edit similarities.     

Robocentric map joining: Improving the consistency of EKF-SLAM

In this paper¹ we study the Extended Kalman Filter approach to simultaneous localization and mapping (EKF-SLAM), describing its known properties and limitations, and concentrate on the filter consistency issue. We show that linearization of the inherent nonlinearities of both the vehicle motion and the sensor models frequently drives the solution of the EKF-SLAM out of consistency, specially in those situations where uncertainty surpasses a certain threshold. We propose a mapping algorithm, Robocentric Map Joining, which improves consistency of the EKF-SLAM algorithm by limiting the level of uncertainty in the continuous evolution of the stochastic map: (1) by building a sequence of independent local maps, and (2) by using a robot centered representation of each local map. Simulations and a large-scale indoor/outdoor experiment validate the proposed approach.