Perceptual Optimization for Point-Based Point Cloud Rendering

Point-based rendering is a common method widely used in point cloud rendering.It realizes rendering by turning the points into the base geometry.The critical step in point-based rendering is to set an appropriate rendering radius for the base geometry,usually calculated using the average Euclidean distance of the N nearest neighboring points to the rendered point.This method effectively reduces the appearance of empty spaces between points in rendering.However,it also causes the problem that the rendering radius of outlier points far away from the central region of the point cloud sequence could be large,which impacts the perceptual quality.To solve the above problem,we propose an algorithm for point-based point cloud rendering through outlier detection to optimize the perceptual quality of rendering.The algorithm determines whether the detected points are outliers using a combination of local and global geometric features.For the detected outliers,the minimum radius is used for rendering.We examine the performance of the proposed method in terms of both objective quality and perceptual quality.The experimental results show that the peak signal-to-noise ratio(PSNR)of the point cloud sequences is improved under all geometric quantization,and the PSNR improvement ratio is more evident in dense point clouds.Specifically,the PSNR of the point cloud sequences is improved by 3.6%on average compared with the original algorithm.The proposed method significantly improves the perceptual quality of the rendered point clouds and the results of ablation studies prove the feasibility and effectiveness of the proposed method.

On Point-Based Haptic Rendering

Haptic rendering is referred to as an approach for complementing graphical model of the virtual object with mechanics- based properties. As a result, when the user interacts with the virtual object through a haptic device, the object can graphically deflect or deform following laws of mechanics. In addition, the user is able to feel the resulting interaction force when interacting with the virtual object. This paper presents a study of defining the levels-of-detail (LOD) in point-based computational mechanics for haptic rendering of objects. The approach uses the description of object as a set of sampled points. In comparison with the finite element method (FEM), point-based approach does not rely on any predefined mesh representation and depends on the point representation of the volume of the object. Different from solving the governing equations of motion representing the entire object based on pre-defined mesh representation which is used in FEM, in point-based modeling approach, the number of points involved in the computation of displacement/deformation can be adaptively defined during the solution cycle. This frame work can offer the implementation of the notion for levels-of-detail techniques for which can be used to tune the haptic rendering environment for in- creased realism and computational efficiency. This paper presents some initial experimental studies in implementing LOD in such environment.

Aggregate Point Cloud Geometric Features for Processing

As 3D acquisition technology develops and 3D sensors become increasingly affordable,large quantities of 3D point cloud data are emerging.How to effectively learn and extract the geometric features from these point clouds has become an urgent problem to be solved.The point cloud geometric information is hidden in disordered,unstructured points,making point cloud analysis a very challenging problem.To address this problem,we propose a novel network framework,called Tree Graph Network(TGNet),which can sample,group,and aggregate local geometric features.Specifically,we construct a Tree Graph by explicit rules,which consists of curves extending in all directions in point cloud feature space,and then aggregate the features of the graph through a cross-attention mechanism.In this way,we incorporate more point cloud geometric structure information into the representation of local geometric features,which makes our network perform better.Our model performs well on several basic point clouds processing tasks such as classification,segmentation,and normal estimation,demonstrating the effectiveness and superiority of our network.Furthermore,we provide ablation experiments and visualizations to better understand our network.

Finite element analysis of slope stability by expanding the mobilized principal stress Mohr's circles-Development, encoding and validation

In recent years,finite element analysis is increasingly being proposed in slope stability problems as a competitive method to traditional limit equilibrium methods(LEMs)which are known for their inherent deficiencies.However,the application of finite element method(FEM)to slope stability as a strength reduction method(SRM)or as finite element limit analysis(FELA)is not always a success for the drawbacks that characterize both methods.To increase the performance of finite element analysis in this problem,a new approach is proposed in this paper.It consists in gradually expanding the mobilized stress Mohr’s circles until the soil failure occurs according to a prescribed non-convergence criterion.The present approach called stress deviator increasing method(SDIM)is considered rigorous for three main reasons.Firstly,it preserves the definition of the factor of safety(FOS)as the ratio of soil shear strength to the mobilized shear stress.Secondly,it maintains the progressive development of shear stress resulting from the increase in the principal stress deviator on the same plane,on which the shear strength takes place.Thirdly,by introducing the concept of equivalent stress loading,the resulting trial stresses are checked against the violation of the actual yield criterion formed with the real strength parameters rather than those reduced by a trial factor.The new numerical procedure was encoded in a Fortran computer code called S^(4)DINA and verified by several examples.Comparisons with other numerical methods such as the SRM,gravity increasing method(GIM)or even FELA by assessing both the FOS and contours of equivalent plastic strains showed promising results.

3D Surface Editing Based on Level-Sets

A novel method which integrates the topological flexibility of the level-set approach and the simplicity of point-sampled surfaces is proposed. The grid structure resulted from the level-set approach not only offers a wide range of powerful surface editing techniques for the point set surface editing, but also facilitates the topological change with ease. With the aid of point-based resampling, the method updates the surface shape of the point-based geometry quickly without worrying about point connectivity at all. The point set surface can also change its topology properly whenever a collision with other parts of itself is detected. The experiment demonstrates their effectiveness on several scanned objects and scan-converted models. Four examples of surface editing operations： smoothing, tapering, deforming, and Boolean operations, are presented.

A two-step surface-based 3D deep learning pipeline for segmentation of intracranial aneurysms

The exact shape of intracranial aneurysms is critical in medical diagnosis and surgical planning.While voxel-based deep learning frameworks have been proposed for this segmentation task,their performance remains limited.In this study,we offer a two-step surface-based deep learning pipeline that achieves significantly better results.Our proposed model takes a surface model of an entire set of principal brain arteries containing aneurysms as input and returns aneurysm surfaces as output.A user first generates a surface model by manually specifying multiple thresholds for time-of-flight magnetic resonance angiography images.The system then samples small surface fragments from the entire set of brain arteries and classifies the surface fragments according to whether aneurysms are present using a point-based deep learning network(PointNet++).Finally,the system applies surface segmentation(SO-Net)to surface fragments containing aneurysms.We conduct a direct comparison of the segmentation performance of our proposed surface-based framework and an existing voxel-based method by counting voxels:our framework achieves a much higher Dice similarity(72%)than the prior approach(46%).

A Fast Approximation Method for Partially Observable Markov Decision Processes

This paper develops a new lower bound method for POMDPs that approximates the update of a belief by the update of its non-zero states.It uses the underlying MDP to explore the optimal reachable state space from initial belief and select actions during value iterations,which significantly accelerates the convergence speed.Also,an algorithm which collects and prunes belief points based on the upper and lower bounds is presented,and experimental results show that it outperforms some of the state-of-art point-based algorithms.

Metrically regular mappings and its application to convergence analysis of a confined Newton-type method for nonsmooth generalized equations

Notion of metrically regular property and certain types of point-based approximations are used for solving the nonsmooth generalized equation f(x)+F(x)?0,where X and Y are Banach spaces,and U is an open subset of X,f:U→Y is a nonsmooth function and F:X■Y is a set-valued mapping with closed graph.We introduce a confined Newton-type method for solving the above nonsmooth generalized equation and analyze the semilocal and local convergence of this method.Specifically,under the point-based approximation of f on U and metrically regular property of f+F,we present quadratic rate of convergence of this method.Furthermore,superlinear rate of convergence of this method is provided under the conditions that f admits p-point-based approximation on U and f+F is metrically regular.An example of nonsmooth functions that have p-point-based approximation is given.Moreover,a numerical experiment is given which illustrates the theoretical result.

View Dependent Sequential Point Trees

Sequential point trees provide the state-of-the-art technique for rendering point models, by re-arranging hierarchical points sequentially according to geometric errors running on GPU for fast rendering. This paper presents a view dependent method to augment sequential point trees by embedding the hierarchical tree structures in the sequential list of hierarchical points. By the method, two kinds of indices are constructed to facilitate the points rendering in an order mostly from near to far and from coarse to fine. As a result, invisible points can be culled view-dependently in high efficiency for hardware acceleration, and at the same time, the advantage of sequential point trees could be still fully taken. Therefore, the new method can run much faster than the conventional sequential point trees, and the acceleration can be highly promoted particularly when the objects possess complex occlusion relationship and viewed closely because invisible points would be in a high percentage of the points at finer levels.