A modified medical image registration

Medical image registration is commonly used in clinical diagnosis, treatment, quality assurance, evaluation of curative efficacy and so on. In this paper, the edges of the original reference and floating images are detected by the B-spline gradient operator and then the binarization images are acquired. By computing the binarization image moments, the centroids are obtained. Also, according to the binarization image coordinates, the rotation angles of the reference and floating images are computed respectively, on the foundation of which the initial values for registering the images are produced. When searching the optimal geometric transformation parameters, the modified peak signal-to-noise ratio (MPSNR) is viewed as the similarity metric between the reference and floating images. At the same time, the simplex method is chosen as multi-parameter optimization one. The experimental results show that, this proposed method has a fairly simple implementation, a low computational load, a fast registration and good registration accuracy. It also can effectively avoid trapping in the local optimum and is adapted to both mono-modality and multi-modality image registrations. Also, the improved iterative closest point algorithm based on acquiring the initial values for registration from the least square method (LICP) is introduced. The experiments reveal that the measure acquiring the initial values for registration from image moments and the least square method (LSM) is feasible and resultful strategy.     

Extreme learning machine with errors in variables

Extreme learning machine (ELM) is widely used in training single-hidden layer feedforward neural networks (SLFNs) because of its good generalization and fast speed. However, most improved ELMs usually discuss the approximation problem for sample data with output noises, not for sample data with noises both in input and output values, i.e., error-in-variable (EIV) model. In this paper, a novel algorithm, called (regularized) TLS-ELM, is proposed to approximate the EIV model based on ELM and total least squares (TLS) method. The proposed TLS-ELM uses the idea of ELM to choose the hidden weights, and applies TLS method to determine the output weights. Furthermore, the perturbation quantities of hidden output matrix and observed values are given simultaneously. Comparison experiments of our proposed TLS-ELM with least square method, TLS method and ELM show that our proposed TLS-ELM has better accuracy and less training time.     

中国虚拟人数据的配准算法

20世纪末出现的基于人体大量信息融合分析的精细化医疗技术,将完整人体的数字化信息和计算机技术相结合,使得基于实际人体信息的模拟和计算成为可能;到目前为止,中国虚拟人的研究刚刚开始。针对中国虚拟人数据的特点,提出了一种图像配准的新方法。根据图像的属性和用途,就配准的各个环节采用合理可行的方法,包括仿射变换、PV插值技术、均方差相似性测量、Powell优化、全局匹配方法。改进了算法流程框架,给出了由参考图像空间到浮动图像空间的逆向几何变换,避免了潜在的“空洞”问题。该算法流程框架将各个部分有机地结合到一起,高效、有序地对中国虚拟人数据图像进行配准。用该配准方法处理中国虚拟人数据,取得了令人满意的效果。     

一类新的RBF 神经网络在非线性系统建模中的应用

提出一种基于输入集分类函数的新的距离度量方法,它与前传回归的正交最小二乘法相结合,不仅可以学习分类超平面的参数,而且可以选择重要的输入节点。这种结构的RBFNN特别适用于非线性动力学系统的辨识（建模）和控制。将改进的RBFNN用于化工中的聚合反应过程建模,结果表明该方法是有效而适用的。     

Comparison of total least squares and instrumental variable methods for parameter estimation of transfer function models

The estimation of the parameters of a transfer function model is considered. Relationships between the total least squares (TLS) and instrumental variable (IV) approaches are outlined. Both methods are able to compute strongly consistent parameter estimates. TLS can be considered as a variation on the IV method where the IV are functions of the time instant and the estimated model parameters. TLS computes strongly consistent estimates of the true model parameters if the outputs and possibly the inputs are independently disturbed by discrete, stationary white noise with zero mean and equal variance. The IV need not be generated. Hence TLS is much simpler to use but more restrictive (IV allows arbitrary noise models) and computationally not so attractive. Next, simulation results are presented comparing the short sample accuracy properties of both methods. When the outputs and possibly the inputs are disturbed by stationary zero mean while noise, TLS outperforms the ordinary IV methods. The accuracy becomes comparable by extending the IV sufficiently. The superiority of TLS is particularly clear in cases where the zeros of the polynomial operating on the outputs are close to the unit circle or where both the inputs and outputs are noisy.     

Geodesics, Parallel Transport & One-Parameter Subgroups for Diffeomorphic Image Registration

Computational anatomy aims at developing models to understand the anatomical variability of organs and tissues. A widely used and validated instrument for comparing the anatomy in medical images is non-linear diffeomorphic registration which is based on a rich mathematical background. For instance, the “large deformation diffeomorphic metric mapping” (LDDMM) framework defines a Riemannian setting by providing a right invariant metric on the tangent spaces, and solves the registration problem by computing geodesics parametrized by time-varying velocity fields. A simpler alternative based on stationary velocity fields (SVF) has been proposed, using the one-parameter subgroups from Lie groups theory. In spite of its better computational efficiency, the geometrical setting of the SVF is more vague, especially regarding the relationship between one-parameter subgroups and geodesics. In this work, we detail the properties of finite dimensional Lie groups that highlight the geometric foundations of one-parameter subgroups. We show that one can define a proper underlying geometric structure (an affine manifold) based on the canonical Cartan connections, for which one-parameter subgroups and their translations are geodesics. This geometric structure is perfectly compatible with all the group operations (left, right composition and inversion), contrarily to left- (or right-) invariant Riemannian metrics. Moreover, we derive closed-form expressions for the parallel transport. Then, we investigate the generalization of such properties to infinite dimensional Lie groups. We suggest that some of the theoretical objections might actually be ruled out by the practical implementation of both the LDDMM and the SVF frameworks for image registration. This leads us to a more practical study comparing the parameterization (initial velocity field) of metric and Cartan geodesics in the specific optimization context of longitudinal and inter-subject image registration.Our experimental results suggests that stationarity is a good approximation for longitudinal deformations, while metric geodesics notably differ from stationary ones for inter-subject registration, which involves much larger and non-physical deformations. Then, we turn to the practical comparison of five parallel transport techniques along one-parameter subgroups. Our results point out the fundamental role played by the numerical implementation, which may hide the theoretical differences between the different schemes. Interestingly, even if the parallel transport generally depends on the path used, an experiment comparing the Cartan parallel transport along the one-parameter subgroup and the LDDMM (metric) geodesics from inter-subject registration suggests that our parallel transport methods are not so sensitive to the path.     

A Self-Stabilizing Neural Algorithm for Total Least Squares Filtering

A neural approach for solving the total least square (TLS) problem is presented in the paper. It is based on a linear neuron with a self-stabilizing neural algorithm, capable of resolving the TLS problem present in the parameter estimation of an adaptive FIR filters for system identification, where noisy errors affect not only the observation vector but also the data matrix. The learning rule is analyzed mathematically and the condition to guarantee the stability of algorithm is educed. The computer simulations are given to illustrate that the neural approach is self-stabilizing and considerably outperforms the existing TLS methods when a larger learning factor is used or the signal-noise-rate is lower.     

Simplifying fuzzy rule-based models using orthogonal transformationmethods

An important issue in fuzzy-rule-based modeling is how to select a set of important fuzzy rules from a given rule base. Even though it is conceivable that removal of redundant or less important fuzzy rules from the rule base can result in a compact fuzzy model with better generalizing ability, the decision as to which rules are redundant or less important is not an easy exercise. In this paper, we introduce several orthogonal transformation-based methods that provide new or alternative tools for rule selection. These methods include an orthogonal least squares (OLS) method, an eigenvalue decomposition (ED) method, a singular value decomposition and QR with column pivoting (SVD-QR) method, a total least squares (TLS) method, and a direct singular value decomposition (D-SVD) method. A common attribute of these methods is that they all work on a firing strength matrix and employ some measure index to detect the rules that should be retained and eliminated. We show the performance of these methods by applying them to solving a nonlinear plant modeling problem. Our conclusions based on analysis and simulation can be used as a guideline for choosing a proper rule selection method for a specific application.     

基于Kalman-OLS的聚丙烯熔融指数软测量

针对聚丙烯装置熔融指数软测量中的非线性和多工况切换操作问题,提出1种基于卡尔曼滤波-正交最小二乘(Kalman-OLS)的非线性自适应软测量方法。通过对聚丙烯装置反应系统进行机理分析,采用正交最小二乘方法(OLS)来拟和辅助变量和熔融指数之间的非线性关系。OLS方法的优化目标函数中同时考虑基于留一法均方误差(LOO MSE)和模型参数的局部正则化(LR),以提高模型的稀疏性和泛化能力。为适应装置多工况操作的现状,进一步提出使用Kalman滤波器对OLS模型参数进行自适应更新。工业数据应用结果表明,提出的Kal-man-OLS方法能够比偏最小二乘(PLS)、OLS方法更准确的预报聚丙烯熔融指数的变化。     

基于梯度场的拼接缝消除方法

提出了一种基于图像梯度场的拼接缝消除方法--GFBSE.在一般的整体变分模型中引入源图像的梯度场,建立一个基于梯度场的能量函数,通过求解非线性偏微分方程以优化该能量函数来实现拼接缝消除.该方法能够在全局消除图像拼接中的颜色不一致,同时能够较好地处理由于图像配准造成的几何结构的错位.最后与当前常用的几种方法进行了理论和实验上的讨论和比较,表明了GFBSE方法的有效性.     

Registration techniques for using imperfect and partially calibrated devices in planar multi-projector displays

Multi-projector displays today are automatically registered, both geometrically and photometrically, using cameras. Existing registration techniques assume pre-calibrated projectors and cameras that are devoid of imperfections such as lens distortion. In practice, however, these devices are usually imperfect and uncalibrated. Registration of each of these devices is often more challenging than the multi-projector display registration itself. To make tiled projection-based displays accessible to a layman user we should allow the use of uncalibrated inexpensive devices that are prone to imperfections. In this paper, we make two important advances in this direction. First, we present a new geometric registration technique that can achieve geometric alignment {\em in the presence of severe projector lens distortion} using a relatively inexpensive low-resolution camera. This is achieved via a closed-form model that relates the projectors to cameras, in planar multi-projector displays, using rational Bezier patches. This enables us to geometrically calibrate a 3000 x 2500 resolution planar multi-projector display made of 3 x 3 array of nine severely distorted projectors using a low resolution (640 x 480) VGA camera. Second, we present a photometric self-calibration technique for a projector-camera pair. This allows us to photometrically calibrate the same display made of nine projectors using a photometrically uncalibrated camera. To the best of our knowledge, this is the first work that allows geometrically imperfect projectors and photometrically uncalibrated cameras in calibrating multi-projector displays.     

一种面向医学图像非刚性配准的多维特征度量方法

医学图像的非刚性配准对于临床的精确诊疗具有重要意义.待配准图像对中目标的大形变和灰度分布呈各向异性给非刚性配准带来困难.本文针对这个问题,提出基于多维特征的联合Renyi α-entropy度量结合全局和局部特征的非刚性配准算法.首先,采用最小距离树构造联合Renyi α-entropy,建立多维特征度量新方法.然后,演绎出新度量准则相对于形变模型参数的梯度解析表达式,采用随机梯度下降法进行参数寻优.最终,将图像的Canny特征和梯度方向特征融入新度量中,实现全局和局部特征相结合的非刚性配准.通过在36对宫颈磁共振（Magnetic resonance,MR）图像上的实验,该方法的配准精度相比较于传统互信息法和互相关系数法有明显提高.这也表明,这种度量新方法能克服因图像局部灰度分布不一致造成的影响,一定程度地减少误匹配,为临床的精确诊疗提供科学依据.     

TDD reciprocity calibration for multi-user massive MIMO systems with iterative coordinate descent多用户大规模MIMO时分双工系统迭代坐标下降互易性校准算法

For massive multiple-input multiple-output (MIMO) antenna systems, time division duplexing (TDD) is preferred since the downlink precoding matrix can be obtained through the uplink channel estimation, thanks to the channel reciprocity. However, the mismatches of the transceiver radio frequency (RF) circuits at both sides of the link make the whole communication channel non-symmetric. This paper extends the total least square (TLS) method to the case of self-calibration, where only the antennas of the access points (APs) are involved to exchange the calibration signals with each other and the feedback from the user equipments (UEs) is not required. Then, the proof of the equivalence between the TLS method and the least square (LS) method is presented. Furthermore, to avoid the eigenvalue decomposition required by these two methods to obtain the calibration coefficients, a novel algorithm named as iterative coordinate descent (ICD) method is proposed. Theoretical analysis and simulation results show that the ICD method significantly reduces the complexity and achieves almost the same performance of the LS method.     

Groupwise geometric and photometric direct image registration

Image registration consists in estimating geometric and photometric transformations that align two images as best as possible. The direct approach consists in minimizing the discrepancy in the intensity or color of the pixels. The inverse compositional algorithm has been recently proposed by Baker et al. for the direct estimation of groupwise geometric transformations. It is efficient in that it performs several computationally expensive calculations at a pre-computation phase. Photometric transformations act on the value of the pixels. They account for effects such as lighting change. Jointly estimating geometric and photometric transformations is thus important for many tasks such as image mosaicing. We propose an algorithm to jointly estimate groupwise geometric and photometric transformations while preserving the efficient pre-computation based design of the original inverse compositional algorithm. It is called the dual inverse compositional algorithm. It uses different approximations than the simultaneous inverse compositional algorithm and handles groupwise geometric and global photometric transformations. Its name stems from the fact that it uses an inverse compositional update rule for both the geometric and the photometric transformations. We demonstrate the proposed algorithm and compare it to previous ones on simulated and real data. This shows clear improvements in computational efficiency and in terms of convergence.     

Nonlinear model structure design and construction using orthogonal least squares and D-optimality design

A very efficient learning algorithm for model subset selection is introduced based on a new composite cost function that simultaneously optimizes the model approximation ability and model robustness and adequacy. The derived model parameters are estimated via forward orthogonal least squares, but the model subset selection cost function includes a D-optimality design criterion that maximizes the determinant of the design matrix of the subset to ensure the model robustness, adequacy, and parsimony of the final model. The proposed approach is based on the forward orthogonal least square (OLS) algorithm, such that new D-optimality-based cost function is constructed based on the orthogonalization process to gain computational advantages and hence to maintain the inherent advantage of computational efficiency associated with the conventional forward OLS approach. Illustrative examples are included to demonstrate the effectiveness of the new approach.     

A method for registration of 3-D shapes

The authors describe a general-purpose, representation-independent method for the accurate and computationally efficient registration of 3-D shapes including free-form curves and surfaces. The method handles the full six degrees of freedom and is based on the iterative closest point (ICP) algorithm, which requires only a procedure to find the closest point on a geometric entity to a given point. The ICP algorithm always converges monotonically to the nearest local minimum of a mean-square distance metric, and the rate of convergence is rapid during the first few iterations. Therefore, given an adequate set of initial rotations and translations for a particular class of objects with a certain level of `shape complexity', one can globally minimize the mean-square distance metric over all six degrees of freedom by testing each initial registration. One important application of this method is to register sensed data from unfixtured rigid objects with an ideal geometric model, prior to shape inspection. Experimental results show the capabilities of the registration algorithm on point sets, curves, and surfaces     

On the effect of measurement errors in regression-adjusted monitoring of multistage manufacturing processes

A critical challenge in multistage process monitoring is the complex relationships between quality characteristics at different stages. A popular method to deal with this problem is regression adjustment in which each quality characteristic is regressed on its preceding quality characteristics and the resulting residual is monitored to detect changes in local variations. However, the performance of this method depends on the accuracy of the regression coefficient estimation. One source of the estimation errors is measurement errors which commonly exist in practice. To provide guidance on the use of regression-adjusted monitoring methods, this study investigates the effect of measurement errors on the bias of regression estimation theoretically and numerically. Two estimators, the ordinary least squares (OLS) estimator and the total least squares (TLS) estimator, are compared, and insights regarding their performance are obtained.     

Urban 3D segmentation and modelling from street view images and LiDAR point clouds

3D urban maps with semantic labels and metric information are not only essential for the next generation robots such autonomous vehicles and city drones, but also help to visualize and augment local environment in mobile user applications. The machine vision challenge is to generate accurate urban maps from existing data with minimal manual annotation. In this work, we propose a novel methodology that takes GPS registered LiDAR (Light Detection And Ranging) point clouds and street view images as inputs and creates semantic labels for the 3D points clouds using a hybrid of rule-based parsing and learning-based labelling that combine point cloud and photometric features. The rule-based parsing boosts segmentation of simple and large structures such as street surfaces and building facades that span almost 75% of the point cloud data. For more complex structures, such as cars, trees and pedestrians, we adopt boosted decision trees that exploit both structure (LiDAR) and photometric (street view) features. We provide qualitative examples of our methodology in 3D visualization where we construct parametric graphical models from labelled data and in 2D image segmentation where 3D labels are back projected to the street view images. In quantitative evaluation we report classification accuracy and computing times and compare results to competing methods with three popular databases: NAVTEQ True, Paris-Rue-Madame and TLS (terrestrial laser scanned) Velodyne.     

小物体3维重建中影像与几何模型的配准

目的 近景摄影测量中的目标几何形状复杂,且拍摄影像的角度变化大,给影像与几何模型的配准带来了困难。传统单幅影像与几何模型配准的做法,由于缺乏自动粗配准的方法,效率相对较低。将多视影像首先统一坐标系的做法,在近景目标的复杂背景下也难以实现。方法 为此,将近景目标置于平面标定板上,利用相机标定的方法同时解算出影像的内外方位元素,实现多视影像坐标系的统一。然后人工选取3组以上同名点,做多视影像与几何模型的绝对定向,得到初始配准参数。最后使用多视影像与几何模型漫反射渲染图之间的归一化互信息作为相似性测度,用Powell非线性优化方法得到配准参数的精确值。结果 实验结果表明,使用标定板可以稳定地获取多视影像的内外方位元素,用绝对定向得到的配准参数进行影像和几何模型的交替显示仍然可以看到明显的裂缝,在经过互信息优化后裂缝现象得到明显改善。结论 多视影像与几何模型配准相比传统单幅影像与几何模型配准,人工选取同名点的工作量大大减少,由于人工选点存在误差,影响绝对定向的精度,使用归一化互信息作为测度进行非线性优化,可以获得更高的精度。     

应用全局最小二乘法辨识AUV运动参数

用系统辨识的方法来估计AUV的水动力参数有很多优点,是当前研究的热点.考虑到AUV运动模型的特点,本文提出用全局最小二乘法(TLS)来辨识AUV水动力参数.首先分析了AUV运动模型,然后讨论了TLS及其与最小二乘法(LS)的区别与联系,提出了辨识模型并且利用仿真数据分别用TLS和LS两种方法进行了辨识,最后分析比较了两种方法的辨识结果得出结论.