基于边缘特征和Keren算法的图像配准

为获得高精确度、高稳定度的岩心配准图像,提出利用Sobel算子提取图像边缘并结合keren算法的图像配准方法。该方法首先提取参考图像和待配准图像边缘,接着对边缘图像进行高斯金字塔分解,最后,利用keren改进算法实现由粗到细的岩心图像配准。实验结果表明该算法与Keren算法相比,在小角度下都有较高的配准精度,大角度旋转情况下该方法绝对误差明显降低,且在大角度旋转下平移参数和角度参数分别可获得0.1个像素以下和0.1度以下的绝对误差精度,提高了岩心配准精度。     

Towards the design and computational characterization of a membrane protein.

The design of a transmembrane four-helix bundle is described. We start with an idealized four-helix bundle geometry, then use statistical information to build a plausible transmembrane bundle. Appropriate residues are chosen using database knowledge on the sequences of membrane helices and loops, then the packing of the bundle core is optimized, and favorable side chain rotamers from rotamer libraries are selected. Next, we use explicit physical knowledge from biomolecular simulation force fields and molecular dynamics simulations to test whether the designed structure is physically possible. These procedures test whether the designed protein will indeed be alpha-helical, well packed and stable over a time scale of several nanoseconds in a realistic lipid bilayer environment. We then test a modeling approach that does not include sophisticated database knowledge about proteins, but rather relies on applying our knowledge of the physics that governs protein motions. This independent validation of the design is based on simulated annealing and restrained molecular dynamics simulation in vacuo, comparable to procedures used to refine NMR and X-ray structures.     

基于二次误差的点云配准算法

提出了一种基于二次误差的特征描述子,该特征描述子具有旋转不变性。通过提取点的二次误差和邻域点二次误差得到两种特征描述子。基于高斯混合模型的点云配准算法层出不穷,主要原因是概率模型在噪声和离群值方面具有更好的鲁棒性,然而该类方法对于尺度较大的旋转表现并不好,为此将二次误差特征描述子作为高斯混合模型的局部特征优化了高斯混合模型较大旋转中的配准效果,并提出基于双特征的配准策略优化了单一特征的缺陷。通过实验与鲁棒的ICP(iterative closest point)以及流行的基于特征的配准算法在配准效率和配准精度方面进行对比,效率是鲁棒性ICP的3～4倍。在大尺度的旋转中提出的算法具有良好的鲁棒性并且优于大多数流行的算法。     

HDRNet: High-Dimensional Regression Network for Point Cloud Registration

Abstract-3D point cloud registration is a crucial topic in the reverse engineering, computer vision and robotics fields. The core of this problem is to estimate a transformation matrix for aligning the source point cloud with a target point cloud. Several learning-based methods have achieved a high performance. However, they are challenged with both partial overlap point clouds and multiscale point clouds, since they use the singular value decomposition (SVD) to find the rotation matrix without fully considering the scale information. Furthermore, previous networks cannot effectively handle the point clouds having large initial rotation angles, which is a common practical case. To address these problems, this paper presents a learning-based point cloud registration network, namely HDRNet, which consists of four stages: local feature extraction, correspondence matrix estimation, feature embedding and fusion and parametric regression. HDRNet is robust to noise and large rotation angles, and can effectively handle the partial overlap and multi-scale point clouds registration. The proposed model is trained on the ModelNet40 dataset, and compared with ICP, SICP, FGR and recent learning-based methods (PCRNet, IDAM, RGMNet and GMCNet) under several settings, including its performance on moving to invisible objects, with higher success rates. To verify the effectiveness and generality of our model, we also further tested our model on the Stanford 3D scanning repository.     

一种权值约束的精确配准算法

针对数据与模型的精确配准问题,提出一种权值约束的配准算法,通过 对配准点施加不同的权值,利用权值约束保证模型重要区域的配准精度。首先,论文基于经 典配准模型,引入权重因子,建立了改进的权值约束的配准模型。针对配准模型的求解问题, 通过对现有SVD-ICP 算法进行适应性改进,提出并研究了带权SVD-ICP(wSVD-ICP)算法, 重点推导了基于wSVD 算法求解旋转矩阵R 和平移矩阵T 的过程。最后,论文利用仿真数 据和实测数据对配准模型进行了验证;计算结果表明,论文所提算法通过对精度要求较高区 域分配高权值进行约束,可有效提升该局部区域的配准精度;同时,可在一定程度上改进整 体配准精度和效率。     

Three-dimensional models of histamine H3 receptor antagonist complexes and their pharmacophore

Molecular modeling was used to analyze the binding mode and activities of histamine H₃ receptor antagonists. A model of the H₃ receptor was constructed through homology modeling methods based on the crystal structure of bovine rhodopsin. Known H₃ antagonists were interactively docked into the putative antagonist binding pocket and the resultant model was subjected to molecular mechanics energy minimization and molecular dynamics simulations which included a continuum model of the lipid bilayer and intra- and extracellular aqueous environments surrounding the transmembrane helices. The transmemebrane helices stayed well embedded in the dielectric slab representing the lipid bilayer and the intra- and extracellular loops remain situated in the aqueous solvent region of the model during molecular dynamics simulations of up to 200 ps in duration. A pharmacophore model was calculated by mapping the features common to three active compounds three-dimensionally in space. The 3D pharmacophore model complements our atomistic receptor/ligand modeling. The H₃ antagonist pharmacophore consists of two protonation sites (i.e. basic centers) connected by a central aromatic ring or hydrophobic region. These two basic sites can simultaneously interact with Asp 114 (3.32) in helix III and a Glu 206 (5.46) in helix V which are believed to be the key residues that histamine interacts with to stabilize the receptor in the active state. The interaction with Glu 206 is consistent with the enhanced activity resulting from the additional basic site. In addition to these two salt bridging interactions, the central region of these antagonists contains a lipophilic group, usually an aromatic ring, that is found to interact with several nearby hydrophobic side chains. The picture of antagonist binding provided by these models is consistent with earlier pharmacophore models for H₃ antagonists with some exceptions.     

基于keren改进配准算法的POCS超分辨率重建

详细介绍了keren亚象素配准算法及其不足,提出了keren算法及其迭代算法的改进算法。该算法基于简化的四参数仿射变换模型而不是传统的刚体变换模型,成功地避免了keren算法因为角度的泰勒级数展开所带来的误差,大大地提高了配准精度。实验仿真结果表明该算法与keren迭代算法相比角度绝对误差显著的降低;平移参数在15°的大角度旋转情况下获得了0.1个象素以下的绝对误差精度,在小角度的情况下获得了0.01个象素以下的绝对误差精度,最后采用POCS方法进行序列图像的高分辨率重建,实验仿真结果表明基于改进配准算法的POCS重建具有良好的配准精度和超分辨率重建效果。     

Three-dimensional models of histamine H3 receptor antagonist complexes and their pharmacophore

Molecular modeling was used to analyze the binding mode and activities of histamine H₃ receptor antagonists. A model of the H₃ receptor was constructed through homology modeling methods based on the crystal structure of bovine rhodopsin. Known H₃ antagonists were interactively docked into the putative antagonist binding pocket and the resultant model was subjected to molecular mechanics energy minimization and molecular dynamics simulations which included a continuum model of the lipid bilayer and intra- and extracellular aqueous environments surrounding the transmembrane helices. The transmemebrane helices stayed well embedded in the dielectric slab representing the lipid bilayer and the intra- and extracellular loops remain situated in the aqueous solvent region of the model during molecular dynamics simulations of up to 200 ps in duration. A pharmacophore model was calculated by mapping the features common to three active compounds three-dimensionally in space. The 3D pharmacophore model complements our atomistic receptor/ligand modeling. The H₃ antagonist pharmacophore consists of two protonation sites (i.e. basic centers) connected by a central aromatic ring or hydrophobic region. These two basic sites can simultaneously interact with Asp 114 (3.32) in helix III and a Glu 206 (5.46) in helix V which are believed to be the key residues that histamine interacts with to stabilize the receptor in the active state. The interaction with Glu 206 is consistent with the enhanced activity resulting from the additional basic site. In addition to these two salt bridging interactions, the central region of these antagonists contains a lipophilic group, usually an aromatic ring, that is found to interact with several nearby hydrophobic side chains. The picture of antagonist binding provided by these models is consistent with earlier pharmacophore models for H₃ antagonists with some exceptions.     

基于边缘几何特征的图像精确匹配方法

提出一组快速高精度计算切线斜率的五点公式,用以估计图像边缘曲线的角度特征,并利用角度直方图估计图像几何变换的旋转参数,实现具有大旋转差异图像间的粗匹配．在进行角度补偿后,利用灰度互相关判据搜索匹配点对,计算出几何变换参数,实现较高精度的旋转和平移校正,最后用松弛迭代法完成图像的精确匹配．与基于小波方向角特征的匹配方法相比,文中方法利用图像中主要的边缘信息实施匹配,具有较好的鲁捧性,可成功实现对各类具有较大相关程度图像间的精确匹配,对图形匹配也具有重要意义．     

Molecular simulation study on concentration effects of rofecoxib with POPC bilayer

The interactions between rofecoxib and POPC (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) bilayer were studied using all-atom molecular dynamics simulation method. Four POPC bilayer systems with different number of rofecoxib molecules were constructed to simulate different drug concentration. The free energy of rofecoxib passing across pure POPC bilayer has two minima (at z ∼1.2 nm or 1.6 nm). As for the high concentration model, the minimum of the free energy profile slightly shifts to the bilayer center. Moreover, the energy change from bulk water to POPC bilayer increases while the central barrier to cross the hydrophobic core of bilayer slightly decreases, suggesting that increasing drug concentration makes it favorable for rofecoxib to partition into the bilayer and easier to pass across bialyer center. Energy analysis show that the stabilization between the selected rofecoxib and other pre-inserted rofecoxib molecule is mainly due to van der Waals interaction energy. The predicted permeability of rofecoxib in high concentration model slightly weakens as compared with low concentration model.     

一种融合纹理的三维图像重建快速实现方法

为了得到完整的三维模型,介绍了一种融合纹理的三维图像重建快速实现方法。通过对不同视角的深度图像的手动粗配准、ICP算法精配准以及全局配准得到这些深度图像的旋转平移矩阵。通过vrippack,三维重建出完整的三维图像,用TextureStitcher对得到的三维图像进行纹理映射,从而实现融合纹理的三维图像的快速重建。文中在论述配准算法主要思想和实现步骤的同时,也用实验验证了方法的可行性与通用性。     

基于面积投影的图像配准方法

针对低分辨率图像之间的配准精度问题,直接影响到超分辨率图像的重建质量.通常图像之间的平移和旋转,采用基于泰勒级数展开的迭代配准算法以及频域配准算法.传统的泰勒级数展开的迭代配准算法的配准精度取决于图像的低阶逼近误差及迭代过程中图像的插值近似运动变换所造成的误差.采用泰勒级数展开的配准算法进行了改进,以面积投影变换来替代原有迭代算法中的图像插值变换,这种图像变换算法更加符合图像的成像原理,仿真结果表明,算法能够有效提高低分辨率图像间平移和旋转角度的配准精度.     

Flexibility of the murine prion protein and its Asp178Asn mutant investigated by molecular dynamics simulations.

Inherited forms of transmissible spongiform encephalopathy, e.g. familial Creutzfeldt-Jakob disease, Gerstmann-Str?ussler-Scheinker syndrome and fatal familial insomnia, segregate with specific point mutations of the prion protein. It has been proposed that the pathologically relevant Asp178Asn (D178N) mutation might destabilize the structure of the prion protein because of the loss of the Arg164-Asp178 salt bridge. Molecular dynamics simulations of the structured C-terminal domain of the murine prion protein and the D178N mutant were performed to investigate this hypothesis. The D178N mutant did not deviate from the NMR conformation more than the wild type on the nanosecond time scale of the simulations. In agreement with CD spectroscopy experiments, no major structural rearrangement could be observed for the D178N mutant, apart from the N-terminal elongation of helix 2. The region of structure around the disulfide bridge deviated the least from the NMR conformation and showed the smallest fluctuations in all simulations in agreement with hydrogen exchange data of the wild type prion protein. Large deviations and flexibility were observed in the segments which are ill-defined in the NMR conformation. Moreover, helix 1 showed an increased degree of mobility, especially at its N-terminal region. The dynamic behavior of the D178N mutant and its minor deviation from the folded conformation suggest that the salt bridge between Arg164 and Asp178 might not be crucial for the stability of the prion protein.     

基于平移域估计的点云全局配准算法

针对迭代最近点(ICP)算法需要两幅点云具有良好的初始位置,否则易陷入局部最优的问题,提出了一种基于平移域估计的点云全局配准算法。首先分别计算数据点云和模型点云的去模糊主方向点云,利用两者平行于坐标轴的包围盒估计平移域范围;其次利用改进的全局ICP算法在估计出的平移域和[-π,π]³的旋转域中进行全局搜索配准。该算法可以根据待配准点云自适应地估计平移域的大小,进行全局自动配准,配准过程中不需要计算点云的特征信息,所需设置的参数少,对点云的初始位置没有要求。实验结果表明,所提算法能够获取全局优化的精确的配准结果,同时提高了全局配准的效率。     

基于立体视觉的3D地形拼接

为了从月球车上的全景相机所捕获的立体图像对重建月球车周围的3D地形,给地面科学家制定科学探测指令提供一个直观的可视化平台,预先展开了基于立体视觉的3D地形重建的研究,主要介绍了在3D地形拼接方面的研究进展。当恢复出多个局部3D地形模型时,首先基于边缘检测和图像匹配技术提取出相邻局部模型之间的公共数据点,然后采用分离旋转变换和平移矢量的策略拟合出相邻模型之间的坐标转换关系,之后就可以将局部模型统一在同一个坐标系下。通过室内和室外多次实验验证了该拼接方案。     

Scale-Free Registrations in 3D: 7 Degrees of Freedom with Fourier Mellin SOFT Transforms

Fourier Mellin SOFT (FMS) as a novel method for global registration of 3D data is presented. It determines the seven degrees of freedom (7-DoF) transformation, i.e., the 6-DoF rigid motion parameters plus 1-DoF scale, between two scans, i.e., two noisy, only partially overlapping views on objects or scenes. It is based on a sequence of the 3D Fourier transform, the Mellin transform and the SO(3) Fourier transform. This combination represents a non-trivial complete 3D extension of the well known Fourier-Mellin registration for 2D images. It is accordingly based on decoupling rotation and scale from translation. First, rotation—which is the main challenge for the extension to 3D data - is tackled with a SO(3) Fourier Transform (SOFT) based on Spherical Harmonics. In a second step, scale is determined via a 3D Mellin transform. Finally, translation is calculated by Phase-Matching. Experiments are presented with simulated data sets for ground truth comparisons and with real world data including object recognition and localization in Magnetic Resonance Tomography (MRT) data, registration of 2.5D RGBD scans from a Microsoft Kinect with a scale-free 3D model generated by Multi-View Vision, and 3D mapping by registration of a sequence of consecutive scans from a low-cost actuated Laser Range Finder. The results show that the method is fast and that it can robustly handle partial overlap, interfering structures, and noise. It is also shown that the method is a very interesting option for 6-DoF registration, i.e., when scale is known.     

Effect of polyelectrolyte size on multilayer conformation and dynamics at different temperatures and salt concentrations

Polyelectrolyte bilayers, which consist of poly-l-lysine (PLL) and hyaluronic acid (HA) were simulated with lipid membranes at different temperatures and ion concentrations. Starting with the sequential deposition of PLL and HA above the membrane surface, PLL and HA become completely mixed, leading to the formation of stable bilayers. PLL/HA bilayers are thicker at higher salt concentration because of weakened electrostatic interactions between PLLs and membrane lipids, in agreement with experiments. This salt effect decreases as PLL size increases. Also, bilayers become thinner at higher temperature because of the increased surface area of membrane. In particular, regardless of temperature and salt concentration, larger PLLs induce thicker bilayers, although larger PLLs have lower diffusivities than do smaller ones. Bilayers with larger PLLs show larger vacancy (more water) inside the bilayer, indicating that larger PLLs are less densely stacked on membrane surface than do smaller ones and thus form the thicker bilayer. These findings show the lower diffusivity of larger polyelectrolytes, which supports the experimental observation regarding the restricted diffusion of large polymers, and also imply the dependence of bilayer thickness on the polymer size.     

基于机器学习和几何变换的实时2D/3D脊椎配准

在图像引导的脊柱手术中,实时高效的2D/3D配准是一项重要且具有挑战性的任务.通常的2D/3D配准一般是将三维图像投影到二维平面,然后进行2D/2D的配准.由于投影空间涉及到3个平移以及3个旋转参数,其投影空间的复杂度为O（n6）,使得配准很难兼具高准确性和高实时性.本文提出了一个结合机器学习与几何变换的2D/3D配准方法,首先,使用统计形状模型对目标脊椎进行建模,并构建了一种新的投影方式,使得6个投影参数中的4个可以使用几何的方法计算出来;接下来利用回归学习的方法学习目标脊椎的形状与投影参数之间的关系;最终,结合学到的关系和几何变换完成配准.本方法的两个姿态参数的平均预测误差为0.84°和0.81°,平均目标配准误差（Mean target registration error,mTRE）为0.87mm,平均配准时间为0.9s.实验结果表明本方法具有很好的实时性和准确性.     

Systematic molecular dynamics searching in a lipid bilayer: application to the glycophorin A and oncogenic ErbB-2 transmembrane domains

Molecular dynamics (MD) simulations of proteins in a lipid bilayer environment are usually undertaken with one or a few starting structures. Here we report a search protocol for systematically exploring the possible interactions in helical bundle transmembrane proteins, a frequently occurring structural motif. The search protocol correctly identifies the experimentally known structure of the dimeric human glycophorin A transmembrane domain as the lowest energy structure among five different models without any prior assumptions, whilst an identical in vacuo search fails to identify the correct structure. The lowest energy structure from the search in a lipid bilayer has a root mean square deviation of 1.1A to the experimental structure. We have applied the same search protocol to the unknown transmembrane structure of the oncogenic mutant ErbB-2 protein, a member of the family of epidermal growth factor receptors. Resulting structures show the role of glutamic acid hydrogen bonding and close helical packing. Water molecules may also play a key role in stabilisation of the transmembrane helix association.     

A frequency domain technique for range data registration

This work introduces an original method for registering pairs of 3D views consisting of range data sets which operates in the frequency domain. The Fourier transform allows the decoupling of the estimate of the rotation parameters from the estimate of the translation parameters, our algorithm exploits this well-known property by suggesting a three-step procedure. The rotation parameters are estimated by the first two steps through convenient representations and projections of the Fourier transforms' magnitudes and the translational displacement is recovered by the third step by means of a standard phase correlation technique after compensating one of the two views for rotation. The performance of the algorithm, which is well-suited for unsupervised registration, is clearly assessed through extensive testing with several objects and shows that good and robust estimates of 3D rigid motion are achievable. Our algorithm can be used as a prealignment tool for more accurate space-domain registration techniques, like the ICP algorithm.