Abstract:Based on the existing Delaunay triangulation method, an algorithm of triangulation growth is presented. This algorithm divides the large-scale point clouds into uniform grids and determines the searching scope self-adaptively. During the process of building a triangulated irregular network (TIN) model, the generated base-lines in groups are grouped, and the close-points are removed dynamically, which improved the speed of reconstructing TIN in large-scale scenes dramatically. By searching the triangular vertices in the scope of the whole data set, this method avoided errors caused by interpolation and the process of stitching between grids. The efficiency and effectiveness of this algorithm are verified by using real world data to build TIN model with large scale LiDAR point clouds.

Abstract:This paper presents an interactive matting approach for efficiently extracting alpha mattes and foreground objects from video sequences. Beginning from user-specified strokes across space and time, the paper formulates their expansions in the video volume as a Laplacian equation, resulting in a coarse alpha matte. It then employs a novel spacetime alpha matting technique that makes use of local statistics and neighboring information and converges to a globally optimal alpha matte in few iterations. Finally, the paper derives a new global cost function to reconstruct the foreground color of the whole video volume, which faithfully preserves the spatio-temporal coherence. The computation in each step can be reformulated as solving a set of linear equations, allowing users to quickly extract high-quality alpha matte and foreground objects, even for data sets with ten million pixels. Experimental results on complex natural video sequences demonstrate the high quality and efficiency of the proposed approach.

Abstract:Image retargeting is the process of adapting images to display terminals with small sizes and different aspect ratios, such as cellular phones and PDAs. This paper presents a novel image retargeting method using saliency-based mesh parameterization. Specifically, this paper formulates retargeting an image to desired size as a constrained mesh parameterization problem which aims at finding a homomorphous target mesh with desired size. This method first constructs a mesh image representation that is consistent with the underlying image structures. To emphasize salient objects and minimize visual distortion, this paper associates image saliency into the mesh and defines image structure as mesh parameterization constraints. Through a stretch-based mesh parameterization process, this paper finally achieves the homomorphous target mesh, which is then used to render the target image by texture mapping. This method generates satisfactory retargeting effects for images with complex background structures. Also it works well for images with multiple salient objects. Experimental results demonstrate the effectiveness of the proposed method.

Abstract:A method that approximates a solid object by object oriented bounding box tree (OBB-Tree) having minimal summed volume outside the object is proposed. First, the outside volume for a single OBB is defined and computed by a hardware-accelerated algorithm. Then, the construction of one OBB-Tree is formulated into a variational approximation. To solve such an approximation, this paper presents an algorithm that minimizes the total outside volume over all OBBs in the same level using the iterative Lloyd clustering and using a variant of iterative MultiGrid among levels. In experiments, comparing against a state-of-the-art alternative, the resulting OBB-Tree is tighter and has better performance in the test of collision detection.

Abstract:The V-system is a new class of complete orthogonal functions system in L2[0,1], which is composed of piecewise kth-order polynomials. There are continuous functions as well as discontinuous functions in V-system. It can be used for signal processing and global expression of a geometric graph group. Moreover, the information of geometric modeling in CAGD can be reconstructed precisely by finite terms of V-system without Gibbs phenomenon, so global feature analysis of the complicated modeling can be implemented. This paper shows that 3-dimension complicated geometric model can be reconstructed by the V-system over triangulated domain. The experiment results indicate that V-system is an effective tool used to reconstruct complicated geometric information with both continuous and discontinuous signals. This is the essential difference among V-system, the classical complete orthogonal system with continuous functions and Walsh and Haar system which include intense discontinuous functions.

Abstract:In this paper, a segment extraction-integrate algorithm based on polygon approximation and finite state machines for on-line Chinese characters recognition (OLCCR) is presented. With this method, the point with the smallest interior angle which is less than the given value is detected and the whole stroke is split into two adjacent curves by this point, which is called as a cut-off point or an inflexion. To each of the two curves, the same step is performed to detect the cut-off points respectively. The same operations are performed iteratively until the smallest interior angle in all the curves is larger than the given threshold value. All the cut-off points and the start-end points compose the stroke and every pair of adjacent points constructs a segment. After segments have been extracted, Finite State Machines is used to check whether the adjacent segments need combination thus redundant segments can be reduced. Experiments proved that this method has the advantages of less computing complexity and better approximating effect than other methods.

Abstract:With powerful computer, high-resolution digital cameras, and sophisticated photo editing software, users can easily manipulate and alter digital images as they wish. Although good forgeries may not be perceptible by human eyes because no visual clues of tampering are left, they may leave some traces of digital tampering that can not be avoided in the media during tampering process. Most digital forgeries employ edge and region smoothing after the contents are manipulated or altered. This paper describes how smoothing introduces disharmony between authentic regions and tampering regions, and then presents a method to automatically detect smoothing regions at any part of an image that indicate possible tampering. The technique works well without any embedded information such as digital watermark.

Abstract:This paper presents a probabilistic method of human action recognition based on manifold learning and Hidden Conditional Random Fields (HCRF). A supervised Neighborhood Preserving Embedding (NPE) is employed for dimensionality reduction by preserving the local neighborhood structure on the data manifold. Most existing approaches to action recognition use a Hidden Markov Model or suitable variant to model actions; a significant limitation of these models is the requirements of conditional independence of observations. In addition, generative models are selected to maximize the likelihood of generating all the examples of a given class and may not uncover the distinctive configuration that sets one class uniquely against others. HCRF relaxes the independence assumption and classifies actions in a discriminative hidden-state formulation. Experimental results on a recent database have demonstrated that this approach can recognize human actions accurately with temporal, intra- and inter-person variations even when noise and other factors such as partial occlusion exist.

Abstract:A grid-based method is presented to extract tetrahedral meshes from the preprocessed volume data, during which the isosurface representing the domain boundary is extracted and the volume inside the domain is tetrahedralized. After the medical volume data is organized into an invisible background grid, a dual method is employed to construct a continuous triangular surface that piecewise linearly approximates the isosurface. To fill the isosurface with tetrahedra, cubes either intersecting with or lying within the isosurface are decomposed by using the precomputed stencils. Finally, Laplacian smoothing is conducted to improve the overall quality of the generated tetrahedral meshes. Regarding that the numerical analysis demands reduced number of elements and accurate geometry near the boundary, adaptive meshing method based on octree-structured grid is also explored. Example of meshing the human distal femur from CT scans is presented, which is applied in the virtual arthroscopic knee surgery.

Abstract:Aiming at describing post probability distribution of the state with high dimensionality by using of a small amount of particles, a sampling in particle filtering for human hand 3D tracking is put forward in this paper. First of all, with the specific human-computer conditions, both cognitive psychology features of operators and motion features of the operators' hands are studied, upon which a novel concept, microstructure of state variable, is proposed. Then, a general method to describe the microstructure of state variable is further discussed. At last, a sampling algorithm based on the microstructure of state variable is put forward. The microstructure of state variable provide an unite and efficient mathematic model upon which the sampling algorithm may avoid sampling mass particles with poor quality. In order to validate validity and performance of sampling algorithm, a great deal of experiments are completed, and just fewer particles, compared with conventional particle filtering, may achieve better tracking precision.

Abstract:With the development of multimedia technology, the use of video has increased in many fields, and captions are frequently inserted into video images to aid the understanding of audience. This paper proposes a robust endpoint detection algorithm for continuous speech in noisy environment, and it can be used in automatic video caption generation systems. In the proposed algorithm, we integrate the widely used energy, zero crossing and entropy to form a new feature, EZE-feature, which possesses advantages while compensating the drawbacks of each individual. Moreover, an adaptive endpoint detection method is proposed which makes the EZE-feature modify its environment parameters by adapting to the strength of background noise. The proposed algorithm has been used in an automatic video caption generation system, and the performance of the algorithm is very well.

Abstract:A process mining approach based on simulated annealing algorithm is proposed, which can be applied to mine complicated structures contained in workflow models, such as non-free choice structures, duplicate tasks and hidden tasks. Experimental results and evaluations of this algorithm are also introduced.

Abstract:An ad hoc network is a collection of wireless mobile nodes dynamically forming a temporary network without the use of any existing network infrastructure or centralized administration. Due to bandwidth constraint and dynamic topology of mobile ad hoc networks, multipath supported routing is a very important research issue. In this paper, we present an entropy-based metric to support stability multipath on-demand routing (SMDR). The key idea of SMDR protocol is to construct the new metric-entropy and select the stability multipath with the help of entropy metric to reduce the number of route reconstruction so as to provide QoS guarantee in the ad hoc network whose topology changes continuously. Simulation results show that, with the proposed multipath routing protocol, packet delivery ratio, end-to-end delay, and routing overhead ratio can be improved in most of cases. It is an available approach to multipath routing decision.

Abstract:As a new learning method, Support Vector Regression (SVR) has good generalization and prediction performance for time series modeling and predicting. In the course of SVR modeling, parameter choosing is very important to the accuracy of models. Aimed at problems in parameter optimization of SVR models, the paper proposes an SVR parameter choosing method for time series prediction, which improves the traditional cross-validation according to the features of time series prediction and sufficiently mines information included in numbered samples on the basis of maintaining the direction characteristic of time series. Furthermore, it is combined with (-weighted SVR in order to get good model parameters. Experimental results over typical time series show the validity of the parameter choosing method of SVR. The method gets good effect applied to time series prediction.

Abstract:This paper proposes a remeshing and multiresolution reconstruction approach based on Laplacian fields. The method starts with the establishment of Laplacian field on an original mesh, and then generates two groups of flow lines with 60° rotation. Finally, a rhombus-dominant mesh is constructed and triangulated into a triangular base-mesh. The method is used again to upsample the original mesh by refining the base-mesh model in order to produce a multiresolution representation. In general, the triangles of the resulting meshes are close to be equilateral. Experiments also show that the method may yield results of high quality.

Abstract:An efficient collision detection method based on separating bounding volume (SBV) is proposed. The positions and shapes of SBVs are determined by the optimal separating support hyper planes of two objects. SBVs not only can efficiently detect the separation of models, but have a high culling ratio when models are intersecting. In order to compute SBVs efficiently, an approximate method using SVM is also put forward and tested. At last in penetration region, a method combined with GPU and SBVs is designed to handle the proximity queries. Experimental results illustrate that SBVs based collision detection algorithm is applicable to exact collision detection for 3D models even without topologies and achieves more efficient and balanced performances in separating, colliding and especially puncturing cases.

Abstract:camera calibration; court model; middle zone; player detection

Abstract:Basing on discretizations of Laplace-Beltrami operator and Gaussian curvature over triangular and quadrilateral meshes and their convergence analyses, this paper proposes in this paper a novel approach for constructing geometric partial differential equation (PDE) Bézier surfaces, using several fourth order geometric flows. Both three-sided and four-sided Bézier surface patches are constructed with G1 boundary constraint conditions. Convergence properties of the proposed method are numerically investigated, which justify that the method is effective and mathematically correct.

Abstract:3D computer aided design (CAD) system based on feature-based solid modeling technique has been widely used for product design recently. At the same time, CAE also plays a more and more important role in product design and finite element method is one of the most popular CAE methods. The models built by CAD systems usually need simplification before finite element analysis according to precision need. How to analysis the degree of simplification is an essential problem in automatic transformation from CAD model to CAE model. Models of different degree of simplification are called Multi-state Model. Considering the need of economizing hardware resource and ensure the appointed correction precision at the same time, this paper proposes a measurement method of the error of different simplification models according to FEM error analysis theory and then validates that by numerical results. Based on the measurement method, the concept of Level of Error is proposed, which provides a new way of getting Multi-state Model in FEM field.

Abstract:Occlusion analysis is a primary function of a mandibular movement simulation system. In this paper, the advantages and drawbacks of concurrent occlusion analysis algorithms are reviewed, and a novel occlusion analysis algorithm based on oriented bounding box trees is proposed by generalizing the Separating Axis Theorem. By means of Heuristic Search, Branch and Bound method and Voronoi Diagram, the algorithm analyzes occlusion on complicated denture models precisely and quickly, which solved the practical problem in the application of VR technologies in dentistry.

Abstract:Concerning the requirements of real-time and accurate collision detection in interactive system, a shared memory parallel collision detection algorithm is proposed. Firstly, the algorithm incorporates the merits of both AABB bounding box and bounding spheres to construct a hybrid bounding representation of arbitrary non-convex polyhedra (S-AABB) for attaining speed, and then uses OpenMP parallel programming model to traversal the built hybrid bounding volume hierarchy, so further accelerates the collision detection. At last, The experimental results have shown that the algorithm is advantageous over other current typical collision detection algorithms such as I-COLLIDE regarding efficiency and accuracy, so can meets the real-time and accurate requirements in complex interactive virtual environment. At the same time, comparing with some parallel collision detection algorithms which include MPI using multi-process, Pipelining using multi-thread and multi-process, this paper has shown that our parallel algorithm based OpenMP is superior in terms of time efficiency, resource consumption and stability.

Abstract:A coarse-to-fine perspective distortion minimization algorithm is proposed for image repairing based on an additional large displacement view (LDV) of the same scene. It works by correcting the perspective distortion in the LDV image, and then utilizing the rectified LDV image to recover the missing areas on the target image. First, under the assumption of a planar scene, the LDV image is globally warped according to a homography to generate the initial distortion correction. Second, a mismatch recognition mechanism detects the remaining distortions in the initially corrected LDV image. They are further relaxed by energy optimization of overlap correspondences with the expectations of color constancy and displacement field smoothness. Third, under the constraints of epipolar geometry, displacement field smoothness and color consistency among the neighboring pixels, the missing pixels are orderly restored according to a specially-defined repairing priority function. Poisson image blending is adopted to eliminate the ghost effect between the repaired region and its surroundings and get the seamless repairing effect. Experimental results demonstrate that this method outperforms recent state-of-the-art image completion algorithms, especially for completing large damaged area with complex structure information.