多分辨率线段提取方法及线段语义分析

线段是一种组成几何体的基本元素,蕴含着非常丰富的几何信息。从图像中提取完整、连续且具有语义信息的线段对恢复场景的几何结构具有重要意义。该文提出了一种多分辨率线段提取方法,并对线段进行语义分析以区分轮廓线段和纹理线段。该方法首先运用多分辨率思想进行线段提取,然后结合深度神经网络技术对线段进行语义分析,最后对线段进行聚类合并得到最终结果。在线段连续性和完整性方面,该文提出的方法与当前常用的线段提取方法相比具有明显优势;在语义分析准确性方面,该文提出的方法在测试集上的像素精度高达 97.82%。     

Perceptual grouping of line features in 3-D space: a model-based framework

In this paper, we propose a novel model-based perceptual grouping algorithm for the line features of 3-D polyhedral objects. Given a 3-D polyhedral model, perceptual grouping is performed to extract a set of 3-D line segments which are geometrically consistent with the 3-D model. Unlike the conventional approaches, grouping is done in 3-D space in a model-based framework. In our unique approach, a decision tree classifier is employed for encoding and retrieving the geometric information of the 3-D model. A Gestalt graph is constructed by classifying input instances into proper Gestalt relations using the decision tree. The Gestalt graph is then decomposed into a few subgraphs, yielding appropriate groups of features. As an application, we suggest a 3-D object recognition system which can be accomplished by selecting a best-matched group. In order to evaluate the performance of the proposed algorithm, experiments are carried out on both synthetic and real scenes.     

基于改进LSD和AP聚类的路径边缘识别策略

起重机金属结构攀爬机器人的路径边缘识别策略分为 3 个步骤。①图像预处理,利 用改进的过颜色算子进行灰度化;②使用基于支持向量机(SVM)最优分类线的方法确定梯度阈 值,并增设主方向角约束,改进线段分割检测(LSD)算法,得到直线段检测图像;③对直线段进 行特征提取,构建聚类数据集,基于数据集动态变化的特点,将基于先验信息的判别模型与近邻 传播(AP)聚类算法相结合,改进 AP 聚类算法,对直线段进行聚类,筛选出构成路径边缘的直线 段,并拟合得到最终的路径边缘线。实验结果表明,相较 AP 聚类和其他聚类算法,改进 AP 聚 类算法的筛选准确率最高;基于改进 LSD 和 AP 聚类的路径边缘识别策略的识别成功率为 96%, 且满足精度和实时性要求。     

Development and evaluation of fast branch-and-bound algorithm for feature matching based on line segments

By extending the previously proposed geometric branch-and-bound algorithm with bounded alignment for point pattern matching, the paper presents the development and evaluation of a new and fast algorithm for image registration based on line segments. Using synthetically generated data sets with randomly distributed line segments and hard test cases with highly symmetric line patterns, as well as real remote sensing images, the developed algorithm is shown to be computationally fast, highly robust, capable of handling severely corrupted data sets with considerable line segment position errors as well as significant fragmented and spurious line segments in the images to be matched.     

一种两点间直线搜索的输电线检测方法

在无人机应用于输电线巡检背景下,为了方便后续的输电线故障检测与分析,提出了一种两点间直线搜索的输电线检测方法。首先,通过方向可控滤波器对无人机图像进行分割。然后,通过所提出的基于圆搜索(CBS)直线段检测方法进行直线段检测。最后,通过所提出的连接算法进行直线段连接。通过在人造图像上的直线段检测实验,证明该方法是一种有效的直线检测方法。随后,通过在实际场景的无人机图像上进行实验,证明该方法是一种有效的输电线检测方法。     

基于窗口Hough变换与阈值分割的矩形识别算法

提出一种基于窗口霍夫变换与阈值分割自动识别图像中的矩形策略： 通过图像窗口霍夫变换,提取霍夫图像的峰值（对应原始图像的线段）,当四个峰值满足某些几何条件时,则检测出矩形;对图像进行阈值分割,将分割结果与霍夫变换的矩形做拟合修正.对不同成像背景和光照环境下图像的集成测试结果表明,本策略能够很好地抑制在多种自然光照不均和拍摄角度造成的干扰.且采用了缩略图计算,降低了逐像素运算的时间复杂度,可满足实时性要求.该技术可运用在实时准确裁剪银行票据目标等各个需要快速识别矩形的工程领域.     

一种基于树结构的立体图像特征提取算法

针对立体图像匹配中的特征检测问题,提出了一种基于树结构的立体图像中边缘点、直线段、交点和凸多边形区域等特征的检测方案。该方法首先基于改进的Canny算子实现立体图像边缘的检测;基于立体图像的边缘梯度信息,文中提出一种分层并行式迭代式链码跟踪方法实现图像中直线段的检测：为提高检测速度并有利于凸多边形区域特征的检测,基于凸多边形区域的几何定义,通过提取直线段交点并采用树结构实现对已检测的相关直线段信息的遍历处理,最终得到树结构表示的凸多边形区域特征。实验证明,所提方法在结构化环境具有很好的检测精度和实时性。     

Estimating motion and structure from correspondences of linesegments between two perspective images

Presents an algorithm for determining 3D motion and structure from correspondences of line segments between two perspective images. To the author's knowledge, this paper is the first investigation of use of line segments in motion and structure from motion. Classical methods use their geometric abstraction, namely straight lines, but then three images are necessary for the motion and structure determination process. In this paper the author shows that it is possible to recover motion from two views when using line segments. The assumption used is that two matched line segments contain the projection of a common part of the corresponding line segment in space, i.e., they overlap. Indeed, this is what the author uses to match line segments between different views. This assumption constrains the possible motion between two views to an open set in motion parameter space. A heuristic, consisting of maximizing the overlap, leads to a unique solution. Both synthetic and real data have been used to test the proposed algorithm, and excellent results have been obtained with real data containing a relatively large set of line segments     

An efficient algorithm for attention-driven image interpretation from segments

In the attention-driven image interpretation process, an image is interpreted as containing several perceptually attended objects as well as the background. The process benefits greatly a content-based image retrieval task with attentively important objects identified and emphasized. An important issue to be addressed in an attention-driven image interpretation is to reconstruct several attentive objects iteratively from the segments of an image by maximizing a global attention function. The object reconstruction is a combinational optimization problem with a complexity of 2^N which is computationally very expensive when the number of segments N is large. In this paper, we formulate the attention-driven image interpretation process by a matrix representation. An efficient algorithm based on the elementary transformation of matrix is proposed to reduce the computational complexity to 3ωN(N-1)²/2, where ω is the number of runs. Experimental results on both the synthetic and real data show a significantly improved processing speed with an acceptable degradation to the accuracy of object formulation.     

On the red/blue spanning tree problem

A geometric spanning tree of a point set S is a tree whose vertex set is S and whose edge set is a set of non-crossing straight line segments with endpoints in S. Given a set of red points and a set of blue points in the plane, the red/blue spanning tree problem is to find a geometric spanning tree for red points and a geometric spanning tree for blue points such that the number of crossing points of the two trees is a minimum. If no three points are collinear, we show that the minimum number of crossing points is completely determined by the number of maximal red (or blue) chains on the convex hull of all red points and blue points. We design an optimal algorithm for constructing a geometric spanning tree of all the red points and a geometric spanning tree of all the blue points with the minimum number of crossing points. If collinear points are allowed, we prove that the problem of deciding whether there exists a geometric spanning path of all the red points and a geometric spanning path of all the blue points without crossing is NP-complete.     

Automatic Road Extraction from Aerial Images

The paper presents a knowledge-based method for automatic road extraction from aerial photography and high-resolution remotely sensed images. The method is based on Marr's theory of vision, which consists of low-level image processing for edge detection and linking, mid-level processing for the formation of road structure, and high-level processing for the recognition of roads. It uses a combined control strategy in which hypotheses are generated in a bottom-up mode and a top-down process is applied to predict the missing road segments. To describe road structures a generalized antiparallel pair is proposed. The hypotheses of road segments are generated based on the knowledge of their geometric and radiometric properties, which are expressed as rules in Prolog. They are verified using part–whole relationships between roads in high-resolution images and roads in low-resolution images and spatial relationships between verified road segments. Some results are presented in this paper.     

Hierarchical generalized Hough transforms and line-segment based generalized Hough transforms

Generalized Hough transforms are fast point pattern matching algorithms which have applications in image processing. Two extensions to generalized Hough transforms are described. The first allows for matching hierarchically organized point patterns. An example concerning image registration using hierarchical point patterns is described. The second extension concerns matching patterns of simple geometric objects, such as line segments, rather than patterns of points.     

Robust and efficient detection of salient convex groups

This paper describes an algorithm that robustly locates salient convex collections of line segments in an image. The algorithm is guaranteed to find all convex sets of line segments in which the length of the gaps between segments is smaller than some fixed proportion of the total length of the lines. This enables the algorithm to find convex groups whose contours are partially occluded or missing due to noise. We give an expected case analysis of the algorithm performance. This demonstrates that salient convexity is unlikely to occur at random, and hence is a strong clue that grouped line segments reflect underlying structure in the scene. We also show that our algorithm run time is O(n ²log(n)+nm), when we wish to find the m most salient groups in an image with n line segments. We support this analysis with experiments on real data, and demonstrate the grouping system as part of a complete recognition system     

Automatic extraction and delineation of single trees from remote sensing data

In this paper, we present a novel approach for the automatic extraction of trees and the delineation of the tree crowns from remote sensing data, and report and evaluate the results obtained with different test data sets. The approach is scale-invariant and is based on co-registered colour-infrared aerial imagery and a digital surface model (DSM). Our primary assumption is that the coarse structure of the crown, if represented at the appropriate level in scale-space, can be approximated with the help of an ellipsoid. The fine structure of the crown is suppressed at this scale level and can be ignored. Our approach is based on a tree model with three geometric parameters (size, circularity and convexity of the tree crown) and one radiometric parameter for the tree vitality. The processing strategy comprises three steps. First, we segment a wide range of scale levels of a pre-processed version of the DSM. In the second step, we select the best hypothesis for a crown from the overlapping segments of all levels based on the tree model. The selection is achieved with the help of fuzzy functions for the tree model parameters. Finally, the crown boundary is refined using active contour models (snakes). The approach was tested with four data sets from different sensors and exhibiting different resolutions. The results are very promising and prove the feasibility of the new approach for automatic tree extraction from remote sensing data. The major part of this work was carried out while both authors worked together at the Institute of Photogrammetry and GeoInformation, University of Hannover.     

Fast dynamic intersection searching in a set of isothetic line segments

We reconsider the (isothetic) line segment intersection searching problem: Given a set S of n horizontal and vertical line segments and a query segment q, find all t segments in S intersecting q. We describe the first dynamic solution for this problem which achieves O(log n + t) query time. This, however, has to be paid by O(n log² n) space requirements and O(log³ n) update time. If segments are defined over a grid of size N × N (the semidynamic case), then the problem can be solved in O(logN + t) query time, O(n log² N) space and O(log² N) update time. The solution is based on the use of segment tree and range tree and the halfobject technique.     

Using B<Superscript>+</Superscript>-trees for processing of line segments in large spatial databases

Points, lines, and regions are the three basic entities for constituting vector-based objects in spatial databases. Many indexing methods (G-tree, K-D-B tree, Quad-tree, PMR-tree, Grid-file, R-tree, and so on) have been widely discussed for handling point or region data. These traditional methods can efficiently organize point or region objects in a space into a hashing or hierarchical directory. They provide efficient access methods to meet the requirement of accurate retrievals. However, two problems are encountered when their techniques are applied to deal with line segments. The first is that representing line segments by means of point or region objects cannot exactly and properly preserve the spatial information about the proximities of line segments. The second problem is derived from the large dead space and overlapping areas in external and internal nodes of the hierarchical directory caused by the use of rectangles to enclose line objects. In this paper, we propose an indexing structure for line segments based on B ⁺ -tree to remedy these two problems. Through the experimental results, we demonstrate that our approach has significant improvement over the storage efficiency. In addition, the retrieval efficiency has also been significantly prompted as compared to the method using R-tree index scheme. These improvements derive mainly from the proposed data processing techniques and the new indexing method.     

利用梯度信息快速提取直线边缘特征

立足于视觉检测系统的实时性需要,提出一种利用梯度信息的快速直线边缘提取方法。该方法首先利用梯度信息和两点确定一条直线进行线段基元的快速定位和扫描;然后对扫描得到的线段基元进行基于几何距离最小化的最佳直线拟合;最后使用端点投影距离的方法对线段基元进行共线性检测,连接共线的线段基元并对连接结果重新进行最佳直线拟合,得到最终的直线边缘特征。实验结果表明:该方法进行直线边缘特征提取的速度比目前文献中最快的Hough变换改进算法提高了1倍左右,适应能力强,可以满足视觉检测系统对直线边缘特征提取的实时性和精度要求。     

Line segment detection using weighted mean shift procedures on a 2D slice sampling strategy

A new line segment detection approach is introduced in this paper for its application in real-time computer vision systems. It has been designed to work unsupervised without any prior knowledge of the imaged scene; hence, it does not require tuning of input parameters. Although many works have been presented on this topic, as far as we know, none of them achieves a trade-off between accuracy and speed as our strategy does. The reduction of the computational cost compared to other fast methods is based on a very efficient sampling strategy that sequentially proposes points on the image that likely belong to line segments. Then, a fast line growing algorithm is applied based on the Bresenham algorithm, which is combined with a modified version of the mean shift algorithm to provide accurate line segments while being robust against noise. The performance of this strategy is tested for a wide variety of images, comparing its results with popular state-of-the-art line segment detection methods. The results show that our proposal outperforms these works considering simultaneously accuracy in the results and processing speed.     

形状感知的绝缘子识别与缺陷诊断

目的 在无人机检测输电线路缺陷的研究中,为提高识别绝缘子的正确率,克服基于颜色来识别绝缘子方法的不足,依据绝缘子串的形状结构特征,研究了一种自底向上感知聚类平行线段的方法。方法 首先将在巡检图像上提取到所有方向的分段划分为6组方向线段,在每一组方向线段中,将线段长度、方向及中心点排列方向一致的线段聚类为平行线组,将平行线组合并,并整理其外接形状,结合输电线路知识模型,可靠识别绝缘子区域。为诊断玻璃绝缘子的掉片缺陷,依据计算出绝缘子的排列方向及片之间距离进行自适应分块,计算每一块的惯性矩均值特征量与惯性矩方差值特征量,依据分块之间特征量相似度来诊断是否存在掉片缺陷。结果 相比基于HSI颜色识别绝缘子的方法,识别绝缘子内部的多平行线段的结构,表现得更稳定,更适用于输电线路巡检。结论 通过无人机巡检采集的输电线路图像,实验结果验证这种方法在复杂背景的条件下能有效识别各种类型绝缘子并能检测绝缘子的掉片缺陷。     

Real- time geometric modeling using models in an actuator space and cartesian space

In this article, three models in two different spaces are compared for use with an image processing system working in real time. A static geometric model of a robot work-cell is held in computer memory as several solid polyhedra. This model is updated as new objects enter or leave the workplace. Similar 2-D slices in joint space, spheres, and simple polyhedra are used to model these objects. The three models are compared for their ability to be updated with new information and for the efficiency of the whole system in accessing data concerning new objects. The system supplies data to a “Path Planner.” The path planner contains a geometric model of the static environment and a robot. The robot machinery structure is modeled as connected cylinders and spheres and the range of motion is quantized. © 1995 John Wiley & Sons, Inc.