Estimating three-dimensional vehicle motion in an outdoor scene using stereo image sequences

We propose a motion estimation system that uses stereo image pairs as the input data. To perform experimental work, we also obtain a sequence of outdoor stereo images taken by two metric cameras. The system consists of four main stages, which are (1) determination of point correspondences on the stereo images, (2) correction of distortions in image coordinates, (3) derivation of 3D point coordinates from 2D correspondences, and (4) estimation of motion parameters based on 3D point correspondences. For the first stage of the system, we use a four-way matching algorithm to obtain matched point on two stereo image pairs at two consecutive time instants (t_i and t_{i + 1}). Since the input data are stereo images taken by cameras, it has two types of distortions, which are (i) film distortion and (ii) lens distortion. These two distortions must be corrected before any process can be applied on the matched points. To accomplish this goal, we use (i) bilinear transform for film distortion correction and (ii) lens formulas for lens distortion correction. After correcting the distortions, the results are 2D coordinates of each matched point that can be used to derive 3D coordinates. However, due to data noise, the calculated 3D coordinates to not usually represent a consistent rigid structure that is suitable for motion estimation; therefore, we suggest a procedure to select good 3D point sets as the input for motion estimation. The procedure exploits two constraints, rigidity between different time instants and uniform point distribution across the object on the image. For the last stage, we use an algorithm to estimate the motion parameters. We also wish to know what is the effect of quantization error on the estimated results; therefore an error analysis based on quantization error is performed on the estimated motion parameters. In order to test our system, eight sets of stereo image pairs are extracted from an outdoor stereo image sequence and used as the input data. The experimental results indicate that the proposed system does provide reasonable estimated motion parameters.     

Computing Cartan subalgebras of Lie algebras

We consider the algorithmic problem of computing Cartan subalgebras in Lie algebras over finite fields and algebraic number fields. We present a deterministic polynomial time algorithm for the case when the ground fieldk is sufficiently large. Our method is based on a solution of a linear algebra problem: the task of finding a locally regular element in a subspace of linear transformations. Also, we give a polynomial time algorithm for restricted Lie algebras over arbitrary finite fields. Both methods require an auxiliary procedure for finding non-nilpotent elements in subalgebras. This problem is also treated. Computational experiences are discussed at the end of the paper.Research supported in part by Hungarian National Foundation for Scientific Research grants T016503 and F4116     

Three dimension reconstruction through measure-based image selection

Three dimension (3D) reconstruction is one of the research focus of computer vision and widely applied in various fields. The main steps of 3D reconstruction include image acquisition, feature point extraction and matching, camera calibration and production of dense 3D scene models. Generally, not all the input images are useful for camera calibration because some images contain similar and redundant visual information. These images can even reduce the calibration accuracy. In this paper, we propose an effective image selection method to improve the accuracy of camera calibration. Then a new 3D reconstruction algorithm is proposed by adding the image selection step to 3D reconstruction. The image selection method uses structure-from-motion algorithm to estimate the position and attitude of each camera, first. Then the contributed value to 3D reconstruction of each image is calculated. Finally, images are selected according to the contributed value of each image and their effects on the contributed values of other images. Experimental results show that our image selection algorithm can improve the accuracy of camera calibration and the 3D reconstruction algorithm proposed in this paper can get better dense 3D models than the normal algorithm without image selection.     

Singular Lagrangian systems and variational constrained mechanics on Lie algebroids

The purpose of this article is to describe Lagrangian mechanics for constrained systems on Lie algebroids, a natural framework which covers a wide range of situations (systems on Lie groups, quotients by the action of a Lie group, standard tangent bundles …). In particular, we are interested in two cases: singular Lagrangian systems and vakonomic mechanics (variational constrained mechanics). Several examples illustrate the interest of these developments.     

Recognition for 3‐D objects based on the determination of the position and the orientation from a noisy single perspective view

Abstract

Recognizing 3‐D objects from 2‐D images has been an important area of research in computer vision. In this paper, an efficient method is developed to determine the position and the orientation of a 3‐D polyhedron in a noisy image. This method assumes that the direction of a straight line on an image can be determined from noisy data. A set of linear equations are used to determine the orientation of an object. In cases where there are not enough straight lines of an image presented, the constraint equations can be introduced and the iterative process is applied. However the iterative process is simple and the divergence is not a problem. After the orientation is determined, the straight lines, corresponding to the visible edges of the object, are used to determine the intersection points. From the intersections and the corresponding corners of the polyhedron, another set of linear equations are used to obtain the position of the object.

If the sum of the perpendicular distances, between the endpoints of the image projection of each model line and the corresponding image line, divided by the number of end‐points is less than a threshold, then the matching process is passed. It is shown, in this paper, that we cannot distinguish between geometrically similar objects without texture analysis from a single perspective view. Thus, the threshold, ?_i , which is determined by a scheme developed here, of model i is used to check whether the candidate model is accepted. Finally, all the concepts, presented in this paper, are illustrated through several examples given in this paper.     

Depth extraction of three-dimensional objects using block matching for slice images in synthetic aperture integral imaging

We describe a computational method for depth extraction of three-dimensional (3D) objects using block matching for slice images in synthetic aperture integral imaging (SAII). SAII is capable of providing high-resolution 3D slice images for 3D objects because the picked-up elemental images are high-resolution ones. In the proposed method, the high-resolution elemental images are recorded by moving a camera; a computational reconstruction algorithm based on ray backprojection generates a set of 3D slice images from the recorded elemental images. To extract depth information of the 3D objects, we propose a new block-matching algorithm between a reference elemental image and a set of 3D slice images. The property of the slices images is that the focused areas are the right location for an object, whereas the blurred areas are considered to be empty space; thus, this can extract robust and accurate depth information of the 3D objects. To demonstrate our method, we carry out the preliminary experiments of 3D objects; the results indicate that our method is superior to a conventional method in terms of depth-map quality.     

类圆柱标签等效阶梯柱面展开方法

目的针对半径不等的类圆柱标签的全表面展开问题,提出类圆柱标签等效阶梯柱面展开方法。方法提取4个视角下图像中的类圆柱标签最小外接矩形,根据提取结果建立位姿估计模型。基于微积分的化曲为直的思想,结合标签的位姿和各视角的轮廓,建立等效阶梯的类圆柱标签3D点云。利用双线性插值法对点云进行渲染,并通过NCC算法进行图像拼接,最终实现半径不等的类圆柱标签全表面展开。结果类圆柱棋盘格标签仿真模型与某品牌口香糖标签全表面展开实验表明,文中提出的算法较传统算法具有较高的精度,针对1号模型,在x轴、y轴方向上的方差分别为1.37像素和0.58像素。结论类圆柱标签等效阶梯柱面展开方法可以有效地实现类圆柱标签全表面展开,为后续的标签检测提供基础。     

结合颜色不变量和ORB特征的图像匹配算法

目的针对现有图像匹配算法因忽略图像颜色信息,使算法对光照变化的鲁棒性较差,容易造成图像误匹配的情况进行改善。方法颜色信息是区分不同图像的重要特征,通过提取彩色图像颜色不变量的方法获得图像的颜色信息,并将颜色不变量与ORB特征描述子相结合,形成一种C-ORB算法,C-ORB算法在保留ORB算法优点的基础上,能够将图像的色彩信息应用到图像匹配过程当中。结果实验结果表明,结合了颜色不变量的ORB算法在保留了ORB算法运算速度快的基础上,能够在不同光照强度、不同光照方向下,获得更多的图像细节,匹配更为准确。结论在移动手机、平板电脑等移动终端设备进行印刷品识别与匹配时,能够在保证识别图像识别速度的基础上进一步减少图像的误匹配。     

Interest points localization for brain image using landmark‐annotated atlas

The localization of clinically important points in brain images is crucial for many neurological studies. Conventional manual landmark annotation requires expertise and is often time‐consuming. In this work, we propose an automatic approach for interest point localization in brain image using landmark‐annotated atlas (LAA). The landmark detection procedure is formulated as a problem of finding corresponding points of the atlas. The LAA is constructed from a set of brain images with clinically relevant landmarks annotated. It provides not only the spatial information of the interest points of the brain but also the optimal features for landmark detection through a learning process. Evaluation was performed on 3D magnetic resonance (MR) data using cross‐validation. Obtained results demonstrate that the proposed method achieves the accuracy of ～ 2 mm, which outperforms the traditional methods such as block matching technique and direct image registration. © 2012 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 22, 145–152, 2012     

A study of the feature‐based digital image stabilization system

Abstract

This article presents a digital image stabilization scheme that uses image processing techniques to compensate for undesirable image jitter due to vehicle or platform vibration to obtain a stabilized image display. The developed digital image stabilization system is based on the image combination approach combined with advanced image selection, feature detection, feature matching and pre‐rotation processing to produce crisp images. Using the Sobel edge detector, we calculate the magnitude of edge response as the judgment criteria for image transformation. To demonstrate the validity of the proposed technique, experimental results are given for some synthetic images.     

3D photon counting integral imaging with unknown sensor positions

Photon counting techniques have been introduced with integral imaging for three-dimensional (3D) imaging applications. The previous reports in this area assumed a priori knowledge of exact sensor positions for 3D image reconstruction, which may be difficult to satisfy in certain applications. In this paper, we extend the photon counting 3D imaging system to situations where sensor positions are unknown. To estimate sensor positions in photon counting integral imaging, scene details of photon counting images are needed for image correspondences matching. Therefore, an iterative method based on the total variation maximum a posteriori expectation maximization (MAP-EM) algorithm is used to restore photon counting images. Experimental results are presented to show the feasibility of the method. To the best of our knowledge, this is the first report on 3D photon counting integral imaging with unknown sensor positions.     

Mapping stereo and image matching algorithms onto parallel architectures

In this paper we present parallel implementations of two vision tasks; stereo matching and image matching. Linear features are used as matching primitives. These implementations are performed on a fixed size mesh array and achieve processor-time optimal performance. For stereo matching, we proposeO(Nn ³/P ²) time algorithm on aP ×P processor mesh array, whereN is the number of line segments in one image,n is the number of line segments in a window determined by the object size, andP ⩽n. The sequential algorithm takesO(Nn ³) time. For image matching, a partitioned parallel implementation is developed.O[((nm/P ²) +P)nm] time performance is achieved on aP ×P processor mesh array, whereP ² ⩽nm. This leads to a processor-time optimal solution forP ⩽ (nm)^1/3. This research was supported in part bynsf under grantiri-9145810 and in part bydarpa andafosr contracts F-49260-89-C-0126 and F-49620-90-C-0078.     

Multiview foreground extraction and composition to multiview background using trimap sharing for natural 3D scene generation

Digital matting for extracting foreground objects from an image is an important process to generate special effects in the movie industry and the broadcasting center. Recently, a digital matting algorithm has been developed to create an alpha matte using a well‐focused image generated from multiview images. However, this method could generate only a single‐view alpha matte, even though it used multiple cameras. In this article, we propose a new estimation scheme for multiview alpha mattes by sharing the trimap of the reference view. Furthermore, we use the motion vector to update the trimap for video matting. After we extract foreground objects from all view images, we composite the foreground objects with the corresponding background images captured in the same multiview camera arrangement. Experimental results demonstrate that multiview composite images can generate reasonably natural 3D views through the stereoscopic monitor. © 2010 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 20, 285‐293, 2010     

Noise reduction using mean shift algorithm for estimating 3D shape

Abstract

The technique to estimate the three-dimensional (3D) geometry of an object from a sequence of images obtained at different focus settings is called shape from focus (SFF). In SFF, the measure of focus — sharpness — is the crucial part for final 3D shape estimation. However, it is difficult to compute accurate and precise focus value because of the noise presence during the image acquisition by imaging system. Various noise filters can be employed to tackle this problem, but they also remove the sharpness information in addition to the noise. In this paper, we propose a method based on mean shift algorithm to remove noise introduced by the imaging process while minimising loss of edges. We test the algorithm in the presence of Gaussian noise and impulse noise. Experimental results show that the proposed algorithm based on the mean shift algorithm provides better results than the traditional focus measures in the presence of the above mentioned two types of noise.     

A high-order Lie groups scheme for solving the recovery of external force in nonlinear system

In this paper, we develop the high-order Lie group scheme to estimate a real-time external force exerted on nonlinear system. For estimating a real-time data, the equations of motion and the supplemental data are combined into a set of differential algebraic equations (DAEs), which is solved by an implicit Lie group differential algebraic equations method (LGDAE) and a Newton iterative algorithm. However, an explicit LGDAE cannot work, and an implicit scheme cannot avoid a lot of iterative numbers and overcome numerical instability under a long time span and noisy-level effect. Because the original implicit LGDAE cannot satisfy the constraint of the cone structure, Lie group and Lie algebra at each time step, solutions highly dependent on ones cannot easily converge by a large time step. Therefore, we develop a high-order explicit LGDAE to avoid iterative numbers and numerical instability, and at the same time combined with higher order sliding modes to obtain real-time external force. The accuracy and efficiency of the novel scheme is validated by comparing the estimation results with the previous literature.     

Dynamic foreground detection based on improved Codebook model

Foreground detection method based on improved Codebook algorithm is discussed in this paper: first of all, transform RGB colour space into YCbCr colour space to make chromaticity convergent, and cope with illumination changing. Second, search for the CodeWords matching with the pixel values of the YCbCr to update the algorithm, so as to ensure the foreground detection effect. Then apply random abandon value method to delete the CodeWords that have not been accessed for a long time, reduce memory consumption and improve processing speed. Finally, conduct experiments on the infrared imagery and the colour image, respectively, to make foreground detection. The results demonstrate that the model and algorithm presented in this paper can make better foreground detection and reduce memory consumption. Meanwhile, compare this method with other algorithms to prove its advantages.     

The Lie-Group Shooting Method for Solving Classical Blasius Flat-Plate Problem

In this paper, we propose a Lie-group shooting method to deal with the classical Blasius flat-plate problem and to find unknown initial conditions. The pivotal point is based on the erection of a one-step Lie group element$\mathbf G(T) and the formation of a generalized mid-point Lie group element$\mathbf G(r). Then, by imposing G(T) = G(r) we can derive some algebraic equations to recover the missing initial conditions. It is the first time that we can apply the Lie-group shooting method to solve the classical Blasius flat-plate problem. Numerical examples are worked out to persuade that the novel approach has better efficiency and accuracy with a fast convergence speed by searching a suitable r ∈(0, 1) with the minimum norm to fit the targets.     

基于Harris与Sift算法的图像匹配方法

采用 Harris 角点检测算法进行图像特征检测, 采用 Sift 算法中的特征描述方法进行图像特征描述, 之后将图像特征点划分为多对多匹配对, 根据特征描述值的支持强度不同建立精确的一对一匹配关系. 该算法有效地避免了图像特征分布均匀时的 Sift 匹配效率较低的问题.