Search

This paper describes a new method for removing geometric distortion in images of submerged objects observed from outside shallow water. We focus on the problem of analyzing video sequences when the water surface is disturbed by waves. The water waves will affect the appearance of the individual video frames such that no single frame is completely free of geometric distortion. This suggests that, in principle, it is possible to perform a selection of a set of low distortion sub-regions from each video frame and combine them to form a single undistorted image of the observed object. The novel contribution in this paper is to use a multi-stage clustering algorithm combined with frequency domain measurements that allow us to select the best set of undistorted sub-regions of each frame in the video sequence. We evaluate the new algorithm on video sequences created both in our laboratory, as well as in natural environments. Results show that our algorithm is effective in removing distortion caused by water motion. [ABSTRACT FROM AUTHOR]

In this paper, a complete theory of steerable filters is presented which shows that quadrature filters are only a special case of steerable filters. Although there has been a large number of approaches dealing with the theory of steerable filters, none of these gives a complete theory with respect to the transformation groups which deform the filter kernel. Michaelis and Sommer (Michaelis and Sommer, 1995) and Hel-Or and Teo (Teo and Hel-Or, 1996; Teo and Hel-Or, 1998) were the first ones who gave a theoretical justification for steerability based on Lie group theory. But the approach of Michaelis and Sommer considers only Abelian Lie groups. Although the approach of Hel-Or and Teo considers all Lie groups, their method for generating the basis functions may fail as shown in this paper. We extend these steerable approaches to arbitrary Lie groups, like the important case of the rotation group SO(3) in three dimensions. Quadrature filters serve for computing the local energy and local phase of a signal. Whereas for the one dimensional case quadrature filters are theoretically well founded, this is not the case for higher dimensional signal spaces. The monogenic signal (Felsberg and Sommer, 2001) based on the Riesz transformation has been shown to be a rotational invariant generalization of the analytic signal. A further generalization of the monogenic signal, the 2D rotational invariant quadrature filter (Köthe, 2003), has been shown to capture richer structures in images as the monogenic signal. We present a generalization of the rotational invariant quadrature filter based on our steerable theory. Our approach includes the important case of 3D rotational invariant quadrature filters but it is not limited to any signal dimension and includes all transformation groups that own a unitary group representation. [ABSTRACT FROM AUTHOR]

Painterly animation is a method that expresses images with a hand-painted appearance from a video, and the most crucial element for it is the coherence between frames. A motion map generation is proposed in this paper as a resolution to the issue of maintaining the coherence in the brush strokes between the frames. A motion map refers to the range of motion calculated by their magnitudes and directions between the frames with the edge of the previous frame as a point of reference. The different methods of motion estimation used in this paper include the optical flow method and the block-based method, and the method that yielded the biggest PSNR using the motion information (the directions and magnitudes) acquired by various methods of motion estimation has been chosen as the final motion information to form a motion map. The created motion map determined the part of the frame that should be re-painted. In order to maintain the temporal coherence, the motion information was applied to only the strong edges that determine the directions of the brush strokes. Also, this paper sought to reduce the flickering phenomenon between the frames by using the multiple exposure method and the difference map created by the difference between images of the source and the canvas. Maintenance of the coherence in the direction of the brush strokes was also attempted by a local gradient interpolation in an attempt to maintain the structural coherence. [ABSTRACT FROM AUTHOR]

Obesity and lack of physical fitness are increasingly common in adults as well as children and can negatively affect health. Regular physical activity, such as jogging or training in a fitness center, is recommended by physiologists to fight obesity and improve one's fitness, but usually requires considerable motivation. Recently, researchers as well as companies have proposed a few fitness games, i.e. videogames where users play by performing physical exercises, in which game elements (such as graphics and gameplay) are used to encourage people to exercise regularly. This paper proposes a fitness game system which aims at combining arcade-style game graphics, physiological sensors (e.g. heart rate monitor, 3D accelerometer), and an adaptation engine. The adaptation engine considers personal information provided by the user (e.g., age and gender), her current heart rate and movements, and information collected during previous game sessions to adjust the required intensity of physical exercises through context-aware and user-adaptive dynamic adaptations of graphics and gameplay. Besides describing the general system, the paper presents two implemented games and a preliminary user evaluation, which also led us to introduce in the system a 3D virtual human. [ABSTRACT FROM AUTHOR]

We introduce in this paper an interface for modeling botanical trees that we term a sketch-and-spray interface. In using the interface, a user would sketch out various types of strokes to construct the main structures of a tree. A user could use a branch stroke to draw a single branch, a leaf stroke to draw a single leaf or a stem of leaves, or a copy stroke to draw a whole branch together with all its sub-branches and leaves. A user could also use various editing strokes to modify the tree structure. Finally, a user could spray particles upon it to ‘grow' the overall shape of the tree. We illustrate the utility of our interface using a number of examples throughout this paper. [ABSTRACT FROM AUTHOR]

In this paper we present a method for fast energy minimization of virtual garments. Our method is based upon the idea of multi-resolution particle system. When garments are approximately positioned around a virtual character, their spring energy may be high, which will cause instability or at least long execution time of the simulation. An energy minimization algorithm is needed; if a fixed resolution is used, it will require many iterations to reduce its energy. Even though the complexity of each iteration is O(n), with a high resolution mass-spring system, this minimization process can take a whole day. The hierarchical method presented in this paper is used to reduce significantly the execution time of the minimization process. The garments are firstly discretized in several resolutions. Once the lowest resolution particles system is minimized (in a short time), a higher resolution model is derived, then minimized. The procedure is iterated up to the highest resolution. But at this stage, the energy to minimize is already much lower so that minimization takes a reasonable time. [ABSTRACT FROM AUTHOR]

This paper describes a novel discrimination method of lung cancers based on statistical analysis of thoracic computed tomography (CT) scans. Our previous Computer-Aided Diagnosis (CAD) system can detect lung cancers from CT scans, but, at the same time, yields many false positives. In order to reduce the false positives, the method proposed in the present paper uses a relationship between lung cancers, false positives and image information on CT scans. The trend of variation of the relationships is acquired through statistical analysis of a set of CT scans prepared for training. In testing, by use of the trend, the method predicts the appearance of lung cancers and false positives in a CT scan, and improves the accuracy of the previous CAD system by modifying the system's output based on the prediction. The method is applied to 218 actual thoracic CT scans with 386 actual lung cancers. Receiver operating characteristic (ROC) analysis is used to evaluate the results. The area under the ROC curve (Az) is statistically significantly improved from 0.918 to 0.931. [ABSTRACT FROM AUTHOR]

This paper describes expression space generation technology that enables animators to control the expressions of 3-dimensional avatars in real-time by selecting a series of expressions from facial expression space. In this system, approximately 2400 facial expression frames are used to generate facial expression space. In this paper, distance matrixes that present distances between facial characteristic points are used to show the state of an expression. The set of these distance matrixes is defined as facial expression space. However, this facial expression space is not space that can be transferred to one space or another in a straight line, when one expression changes to another. In this technology, the route for moving from one expression to another is approximately inferred from captured facial expression data. First, it is assumed that two expressions are close to each other when the distance between distance matrixes that show facial expression states is below a certain value. When two random facial expression states are connected with the set of a series of adjacent expressions, it is assumed that there is a route between the two expressions. It is further assumed that the shortest path between two facial expressions is the path when one expression moves to the other expression. Dynamic programming is used to find the shortest path between two facial expressions. The facial expression space, which is the set of these distance matrixes, is multidimensional space. The facial expression control of 3-dimensional avatars is carried out in real-time when animators navigate through facial expression space. In order to assist this task, multidimensional scaling is used for visualization in 2-dimensional space, and animators are told to control facial expressions when using this system. This paper evaluates the results of the experiment. [ABSTRACT FROM AUTHOR]

In this paper, we propose a new gaze estimation method by analyzing the cornea surface model which is estimated through three dimensional analysis of human eye in HMD (Head Mounted Display) environments. This paper has four advantages over previous works. First, in order to obtain accurate gaze position, we use a cornea sphere model based on Gullstrand eye scheme. Second, we calculate the 3D position of the cornea sphere and a gaze vector by using a camera, three collimated IR-LEDs and one illuminated IR-LED. Third, three coordinates such as camera, monitor and eye coordinates are unified, which can simplify the complex 3D converting calculation and allow for calculation of the 3D eye position and gaze position on a HMD monitor. Fourth, a simple user dependent calibration method is proposed by gazing at one position of HMD monitor based on Kappa compensation. Experimental results showed that the average gaze estimation error of the proposed method was 0.89 degrees. [ABSTRACT FROM AUTHOR]

Non-Photorealistic Rendering (NPR) can offer increased concentration and familiarity to the watcher. For this reason, many media such as movies, games, and commercials currently use the NPR method to deliver information. In this paper, we suggest an interactive system for video cartooning based on the mean-shift segmentation of image and video. In order to solve the problems of time complexity and memory allocation, the conventional problems of video mean-shift segmentation, this paper proposes several techniques such as foreground object based segmentation and sequential segmentation. We also propose the interactive correction technique to get enhanced results. For more cartoonic representation, we used spline curve approximation of segment boundaries in the final rendering results. With our method, we can easily create cartoon rendering output using the video streams such like home video, which can be obtained easily in our daily life. [ABSTRACT FROM AUTHOR]

Scale space interest points capture important photometric and deep structure information of an image. The information content of such points can be made explicit using image reconstruction. In this paper we will consider the problem of combining multiple types of interest points used for image reconstruction. It is shown that ordering the complete set of points by differential (quadratic) TV-norm (which works for single feature types) does not yield optimal results for combined point sets. The paper presents a method to solve this problem using canonical sets of scale space features. Qualitative and quantitative analysis show improved performance over simple ordering of points using the TV-norm. [ABSTRACT FROM AUTHOR]

This paper reports the results obtained by analysing some of the most well-known features used in Computer Vision to describe and classify shapes in an appealing real application. We aim to demonstrate the applicability of shape descriptors to classify muscles on Magnetic Resonance Imaging (MRI). The mechanized classification of ham muscles could help the industries to automate the ripening process for Iberian ham. The excellent classification percentages obtained in our experiments suggest the real viability of the feature vector developed in this paper to recognize and classify muscles. [ABSTRACT FROM AUTHOR]

In traffic sign recognition systems, one of the normal approaches is the identification of the shape of the sign prior to the recognition itself. Normally, the recognition process needs an accurate localization of the sign for a good performance. If we are dealing with triangular, rectangular and circular signs, this means the accurate localization of the vertices of the triangle and the rectangle, or the parameters of the ellipse. In this paper we have developed a system which searches the above mentioned parameters from the signature of the blob using techniques of algebraic-distance minimization. Comparisons with previous works show good improvements in the localization of the shape, especially in the presence of slight occlusions. This work is part of a traffic sign recognition system, and in this paper we focus on the shape localization step. [ABSTRACT FROM AUTHOR]

This paper deals with a compact catadioptric omnidirectional stereovision system based on a single camera and multi-mirrors (at least two mirrors). Many configurations were empirically designed in previous works with the aim to obtain a good 3D reconstruction accuracy. In this paper, we propose to use optimization techniques for omnidirectional catadioptric stereovision design, by using a stochastic local search method in order to find a good sensor (number, relative positions and sizes of mirrors). We explain principles of our approach and provide automatically designed sensors with a number of mirrors from two to nine. We finally simulate the 3D-reconstruction of a real environment modeled under a ray-tracing software with some of these sensors. [ABSTRACT FROM AUTHOR]

This paper addresses the problem of reconstructing non- overlapping transparent and opaque surfaces from multiple view images. The reconstruction is attained through progressive refinement of an initial 3D shape by minimizing the error between the images of the object and the initial 3D shape. The challenge is to simultaneously reconstruct both the transparent and opaque surfaces given only a limited number of images. Any refinement methods can theoretically be applied if analytic relation between pixel value in the training images and vertices position of the initial 3D shape is known. This paper investigates such analytic relations for reconstructing opaque and transparent surfaces. The analytic relation for opaque surface follows diffuse reflection model, whereas for transparent surface follows ray tracing model. However, both relations can be converged for reconstruction both surfaces into texture mapping model. To improve the reconstruction results several strategies including regularization, hierarchical learning, and simulated annealing are investigated. [ABSTRACT FROM AUTHOR]

An essential functionality for advanced driver assistance systems (ADAS) is road segmentation, which directly supports ADAS applications like road departure warning and is an invaluable background segmentation stage for other functionalities as vehicle detection. Unfortunately, road segmentation is far from being trivial since the road is in an outdoor scenario imaged from a mobile platform. For instance, shadows are a relevant problem for segmentation. The usual approaches are ad hoc mechanisms, applied after an initial segmentation step, that try to recover road patches not included as segmented road for being in shadow. In this paper we argue that by using a different feature space to perform the segmentation we can minimize the problem of shadows from the very beginning. Rather than the usual segmentation in a color space we propose segmentation in a shadowless image which is computable in real-time using a color camera. The paper presents comparative results for both asphalted and non-asphalted roads, showing the benefits of the proposal in presence of shadows and vehicles. [ABSTRACT FROM AUTHOR]

This paper outlines a neural network model based on the Fuzzy Cognitive Maps (FCM) framework for solving the automatic image change detection problem. Each pixel in the reference image is assumed to be a node in the network. Each node has associated a fuzzy value, which determines the magnitude of the change. Each fuzzy value is updated by a trade-off between the influences received from the fuzzy values from other neurons and its own fuzzy value. Classical approaches in the literature have been designed assuming that the mutual influences between two nodes are symmetric. The main finding of this paper is the assumption that mutual influences could not be symmetric. This non symmetric relationship can be embedded by the FCM paradigm. The performance of the proposed method is illustrated by comparative analysis against some recent image change detection methods. [ABSTRACT FROM AUTHOR]

This paper proposes an incremental method for feature selection, aimed at identifying attributes in a dataset that allow to buid good classifiers at low computational cost. The basis of the approach is the minimal-redundancy-maximal-relevance (mRMR) framework, which attempts to select features relevant for a given classification task, avoiding redundancy among them. Relevance and redundancy have been popularly defined in terms of information theory concepts. In this paper a modification of the mRMR framework is proposed, based on a more proper quantification of the redundancy among features. Experimental work on discrete-valued datasets shows that classifiers built using features selected by the proposed method are more accurate than the ones obtained using original mRMR features. [ABSTRACT FROM AUTHOR]

Pulmonary embolism (PE) is a very serious condition causing sudden death in about one-third of the cases. Treatment with anti-clotting medications is highly effective but not without complications, while diagnosis has been missed in about 70% of the cases. A major clinical challenge, particularly in an Emergency Room, is to quickly and correctly diagnose patients with PE and then send them on to therapy. Computed tomographic pulmonary angiography (CTPA) has recently emerged as an accurate diagnostic tool for PE, but each CTPA study contains hundreds of CT slices. The accuracy and efficiency of interpreting such a large image data set is complicated by various PE look-alikes and also limited by human factors, such as attention span and eye fatigue. In response to this challenge, in this paper, we present a fast yet effective approach for computer aided detection of pulmonary embolism in CTPA. Our proposed approach is capable of detecting both acute and chronic pulmonary emboli with a distinguished feature of incrementally reporting any detection immediately once becoming available during searching, offering real-time support and achieving 80% sensitivity at 4 false positives. This superior performance is contributed to our novel algorithms (concentration oriented tobogganing and multiple instance classification) introduced in this paper for candidate detection and false positive reduction. [ABSTRACT FROM AUTHOR]

In this paper we describe a new method for quantifying metabolic asymmetry modulo structural hemispheric differences. The study of metabolic asymmetry in Alzheimer's disease (AD) serves as a driving application. The approach is based on anatomical atlas construction by large deformation diffeomorphic metric mapping (LDDMM) first introduced in [1]. Using invariance properties of the LDDMM, we define a structurally symmetric coordinate frame in which metabolic asymmetries between the left and the right hemispheres can be studied. This structurally symmetric coordinate system of each subject provides the correspondence between left and right hemispheric structures in an individual brain. These correspondences are used for measuring metabolic asymmetry modulo structural asymmetry. Again using the atlas construction framework, we build a common symmetric coordinate system of a entire population. The metabolic asymmetry maps of individuals in a population under study are mapped into the common structurally symmetric coordinate frame, allowing for a statistical description of the populations metabolic asymmetry. In this paper we prove certain invariance properties of the LDDMM atlas construction framework that make the definition of structurally symmetric coordinate systems possible. We present results from applying the methodology to images from the Alzheimer's Disease Neuroimaging Initiative (ADNI)[2]. [ABSTRACT FROM AUTHOR]

This paper presents a new method for constructing compact statistical point-based models of ensembles of similar shapes that does not rely on any specific surface parameterization. The method requires very little preprocessing or parameter tuning, and is applicable to a wider range of problems than existing methods, including nonmanifold surfaces and objects of arbitrary topology. The proposed method is to construct a point-based sampling of the shape ensemble that simultaneously maximizes both the geometric accuracy and the statistical simplicity of the model. Surface point samples, which also define the shape-to-shape correspondences, are modeled as sets of dynamic particles that are constrained to lie on a set of implicit surfaces. Sample positions are optimized by gradient descent on an energy function that balances the negative entropy of the distribution on each shape with the positive entropy of the ensemble of shapes. We also extend the method with a curvature-adaptive sampling strategy in order to better approximate the geometry of the objects. This paper presents the formulation; several synthetic examples in two and three dimensions; and an application to the statistical shape analysis of the caudate and hippocampus brain structures from two clinical studies. [ABSTRACT FROM AUTHOR]

In this paper an algorithm for atlas-to-image non-rigid registration based on regional entropy minimization is presented. Tissue class probabilities in the atlas are registered with the intensities in the target image. The novel aspect of the paper consists in using tissue class probability maps that include the three main regions (for the brain, white matter, gray matter and csf) and a further partitioning thereof. For example, gray matter is further subdivided into basal ganglia (each of them defining its own class) and the rest (of gray matter). This guarantees a regional entropy minimization instead of just a global one.In other words, the local labels in the atlas will be adjusted in order to obtain the best explanation for the intensity distribution in the corresponding subregion of the target image. [ABSTRACT FROM AUTHOR]

This paper presents a novel statistical fuzzy-segmentation method for diffusion tensor (DT) images and magnetic resonance (MR) images. Typical fuzzy-segmentation schemes, e.g. those based on fuzzy-C-means (FCM), incorporate Gaussian class models which are inherently biased towards ellipsoidal clusters. Fiber bundles in DT images, however, comprise tensors that can inherently lie on more-complex manifolds. Unlike FCM-based schemes, the proposed method relies on modeling the manifolds underlying the classes by incorporating nonparametric data-driven statistical models. It produces an optimal fuzzy segmentation by maximizing a novel information-theoretic energy in a Markov-random-field framework. For DT images, the paper describes a consistent statistical technique for nonparametric modeling in Riemannian DT spaces that incorporates two very recent works. In this way, the proposed method provides uncertainties in the segmentation decisions, which stem from imaging artifacts including noise, partial voluming, and inhomogeneity. The paper shows results on synthetic and real, DT as well as MR images. [ABSTRACT FROM AUTHOR]

This paper presents a general approach for surface-to-surface registration (S2SR) with the Euclidean metric using signed distance maps. In addition, the method is symmetric such that the registration of a shape A to a shape B is identical to the registration of the shape B to the shape A. The S2SR problem can be approximated by the image registration (IR) problem of the signed distance maps (SDMs) of the surfaces confined to some narrow band. By shrinking the narrow bands around the zero level sets the solution to the IR problem converges towards the S2SR problem. It is our hypothesis that this approach is more robust and less prone to fall into local minima than ordinary surface-to-surface registration. The IR problem is solved using the inverse compositional algorithm. In this paper, a set of 40 pelvic bones of Duroc pigs are registered to each other w.r.t. the Euclidean transformation with both the S2SR approach and iterative closest point approach, and the results are compared. [ABSTRACT FROM AUTHOR]

This paper describes a novel method of extracting moving region boundaries from the frames of an image sequence by filtering information of the first temporal cooccurrence matrix using the corresponding matrix from the next frame. The method described in this paper does not make use of any threshold and it is very robust and efficient with respect to noise. [ABSTRACT FROM AUTHOR]

Since 1987 it is known that the Euclidean shortest path problem is NP-hard. However, if the 3D world is subdivided into cubes, all of the same size, defining obstacles or possible spaces to move in, then the Euclidean shortest path problem has a linear-time solution, if all spaces to move in form a simple cube-curve. The shortest path through a simple cube-curve in the orthogonal 3D grid is a minimum-length polygonal curve (MLP for short). So far only one general and linear (only with respect to measured run times) algorithm, called the rubberband algorithm, was known for an approximative calculation of an MLP. The algorithm is basically defined by moves of vertices along critical edges (i.e., edges in three cubes of the given cube-curve). A proof, that this algorithm always converges to the correct MLP, and if so, then always (provable) in linear time, was still an open problem so far (the authors had successfully treated only a very special case of simple cube-curves before). In a previous paper, the authors also showed that the original rubberband algorithm required a (minor) correction. This paper finally answers the open problem: by a further modification of the corrected rubberband algorithm, it turns into a provable linear-time algorithm for calculating the MLP of any simple cube-curve. The paper also presents an alternative provable linear-time algorithm for the same task, which is based on moving vertices within faces of cubes. For a disticntion, we call the modified original algorithm now the edge-based rubberband algorithm, and the second algorithm is the face-based rubberband algorithm; the time complexity of both is in ${\cal O}(m)$, where m is the number of critical edges of the given simple cube-curve. [ABSTRACT FROM AUTHOR]

This paper extends the concepts of image matching in the non-parametric space and binary distance measures. Matching in the nonparametric domain exhibits many desirable properties at relatively small computation complexity: It concentrates on capturing mutual relation among pixels in a small neighbourhoods rather than bare intensity values, thus improving matching discrimination. It is also more resistive against noise and uneven lighting conditions of the matched images. Last but not least, the matching algorithms operate in the integer domain and can be easily implemented in hardware what benefits in dramatic improvement of their run times. In this paper we extend the concept of nonparametric image transformation into the realm of colour images taking into consideration different colour spaces and different distances defined in these spaces. We propose significant bit reduction for aggregated block matching in the Census domain. We propose also the sparse sampling model for the Census transformation that increase the discriminative power of this representation and allows even further reduction of bits necessary for matching. The presented techniques have been applied to matching of the stereo images but can be employed in any computer vision task that requires comparison of images, such as image registration, object detection and recognition, etc. Presented experiments exhibit interesting properties of the described techniques. Keywords: image matching, Census, binary distance. [ABSTRACT FROM AUTHOR]

Branch indices of points on curves (introduced by Urysohn and Menger) are of basic importance in the mathematical theory of curves, defined in Euclidean space. This paper applies the concept of branch points in the 3D orthogonal grid, motivated by the need to analyze curve-like structures in digital images. These curve-like structures have been derived as 3D skeletons (by means of thinning). This paper discusses approaches of defining branch indices for voxels on 3D skeletons, where the notion of a junction will play a crucial role. We illustrate the potentials of using junctions in 3D image analysis based on a recent project of analyzing the distribution of astrocytes in human brain tissue. Keywords: 3D skeletons, 3D curve analysis, branch nodes, branch index, thinning, medical image analysis, astrocytes. [ABSTRACT FROM AUTHOR]

Construction of geometrical objects by origami, the Japanese traditional art of paper folding, is enjoyable and intriguing. It attracted the minds of artists, mathematicians and computer scientists for many centuries. Origami will become a more rigorous, effective and enjoyable art if the origami constructions can be visualized on the computer and the correctness of the constructions can be automatically proved by an algorithm. We call the methodology of visualizing and automatically proving origami constructions computational origami. As a non-trivial example, in this paper, we visualize a construction of a regular heptagon by origami and automatically prove the correctness of the construction. [ABSTRACT FROM AUTHOR]

Rebuilding three-dimensional objects represented by a set of points is a classical problem in computer graphics. Multiple applications like medical imaging or industrial techniques require finding shape from scattered data. Therefore, the reconstruction of a set of points that represents a shape has been widely studied, depending on data source and reconstruction's objectives. This purpose of this paper is to provide an automatic reconstruction from an unorganized cloud describing an unknown shape in order to provide a solution that will allow to compute the object's volume and to deform it with constant volume. The main idea in this paper consists in filling the object's interior with an equipotential surface resulting of the fusion of potential field primitives also called metaballs or blobs. Nevertheless, contrary to most of usual rebuilding methods based on implicit primitives blending, we do not compute any medial axis to set the primary objects. Indeed, a fast voxelization is used to find a summary contour from the discrete shape and to determine interior areas. Then, the positioning of implicit primitives rely on a multilayer system. Finally, a controlled fusion of the isosurfaces guarantees the lack of any holes and a respectful contour of the original object, such that we obtain a complete shape filling. Keywords: Reconstruction, Cloud of dots, Implicit surfaces, Blobs, Voxelization. [ABSTRACT FROM AUTHOR]

In this paper, we want to exploit the knowledge obtained from those detected objects which are incorporated into the background model since they cease their movement. These motionless foreground objects should be handled in security domains such as video surveillance. This paper uses an adaptive background modelling algorithm for moving-object detection. Those detected objects which present no motion are identified and added into the background model, so that they will be part of the new background. Such motionless agents are included for further appearance analysis and agent categorization. [ABSTRACT FROM AUTHOR]

This paper presents a new scheme for synthesizing hand motion to grasp various objects. Hand motion has a variety of expression with its high degrees of freedom and the functional motions are especially complex and difficult to synthesize. This paper focuses on a fact that actual grasp motion varies depending on the features of the object including its size and shape, which give important clues for reproducing a proper hand motion to manipulate them. Based on this idea, we propose a scheme to sample grasp motions and synthesize the whole hand motion including approach to the object, preparation of the hand shape and grab motion. Synthesized animation demonstrated a potential for easily designing functional motions for hand animation. Keywords: Hand motion, grasp, motion capture, animation. [ABSTRACT FROM AUTHOR]

This paper addresses the problem of probabilistic modelling of human motion by combining several 2D views. This method takes advantage of 3D information avoiding the use of a complex 3D model. Considering that the main disadvantage of 2D models is their restriction to the camera angle, a solution to this limitation is proposed in this paper. A multi-view Gaussian Mixture Model (GMM) is therefore fitted to a feature space made of Shapes and Stick figures manually labelled. Temporal and spatial constraints are considered to build a probabilistic transition matrix. During the fitting, this matrix limits the feature space only to the most probable models from the GMM. Preliminary results have demonstrated the ability of this approach to adequately estimate postures independently of the direction of motion during the sequence. [ABSTRACT FROM AUTHOR]

This paper introduces AcouMotion as a new hard-/software system for combining human body motion, tangible interfaces and sonification to a closed-loop human computer interface that allows non-visual motor control by using sonification (non-speech auditory displays) as major feedback channel. AcouMotion's main components are (i) a sensor device for measuring motion parameters (ii) a computer simulation to represent the dynamical evolution of a model world, and (iii) a sonification engine which generates an auditory representation of objects and any interactions in the model world. The intended applications of AcouMotion range from new kinds of sport games that can be played without visual displays and therefore may be particularly interesting for people with visual impairment to further applications in data mining, physiotherapy and cognitive research. The first application of AcouMotion presented in this paper is Blindminton, a sport game similar to Badminton which is particularly adapted to the abilities of people with visual impairment. We describe our current system and its state of development, and we present first sound examples for interactive sonification using an early prototype. Finally, we discuss some interesting research directions based on the fact that AcouMotion binds auditory stimuli and body motion, and thus can represent a counterpart to the Eye-tracker device that exploits the binding of visual stimuli and eye-movement in cognitive research. [ABSTRACT FROM AUTHOR]

Simulating realistic human-like figures is still a challenging task when dynamics is involved. For example, making a virtual human jump to a given position requires to control the forces involved in take-off in order to reach a given velocity vector at the beginning of the aerial phase. Several problems are addressed in this paper in order to modify a captured motion while accounting from dynamics. The method exploits a point mass approximation of the body for the Inverse Dynamics stage during the contact phase and later to optimize new trajectories. First, accurate body segment masses are required to have access to external forces thanks to inverse dynamics. Second, those forces have to be adapted to make the resulting center of mass trajectory verify new constraints (such as reaching a given point at a given time). This paper also proposes a new formalism to encode force depending on time in contact phases (called impulse). Whereas classical biomechanical analyzes focus only on the peak of forces and on the contact phase duration, our formalism provides new data to characterize the shape of an impulse. Keywords: take-off, optimization, dynamic control, constraints. [ABSTRACT FROM AUTHOR]

Hitherto, one major challenge to sign language recognition is how to develop approaches that scale well with increasing vocabulary size. In large vocabulary speech recognition realm, it is effective to use phonemes instead of words as the basic units. This idea can be used in large vocabulary Sign Language recognition, too. In this paper, Etyma are defined to be the smallest unit in a sign language, that is, a unit that has some meaning and distinguishes one sign from the others. They can be seen as phonemes in Sign Language. Two approaches to large vocabulary Chinese Sign Languagerecognition are discussed in this paper. One uses etyma and the other uses whole signs as the basic units. Two CyberGloves and a Pohelmus 3-D tracker with three receivers positioned on the wrist of CyberGlove and the back are used as input device. Etymon- and word- based recognition systems are introduced, which are designed to recognize 2439 etyma and 5100 signs. And then the experimental results of these two systems are given and analyzed. [ABSTRACT FROM AUTHOR]

This paper applies singularity theory of mappings of surfaces to 3-space and the generic transitions occurring in their deformations to develop algorithms for continuously and robustly tracking the intersection curves of two deforming parametric spline surfaces, when the deformation is represented as a family of generalized offset surfaces. This paper presents the mathematical framework, and develops algorithms accordingly, to continuously and robustly track the intersection curves of two deforming parametric surfaces, with the deformation represented as generalized offset vector fields. The set of intersection curves of 2 deforming surfaces over all time is formulated as an implicit 2-manifold $\mathcal{I}$ in the augmented (by time domain) parametric space $\mathbb R^5$. Hyper-planes corresponding to some fixed time instants may touch$\mathcal{I}$ at some isolated transition points, which delineate transition events, i.e., the topological changes to the intersection curves. These transition points are the 0-dimensional solution to a rational system of 5 constraints in 5 variables, and can be computed efficiently and robustly with a rational constraint solver using subdivision and hyper-tangent bounding cones. The actual transition events are computed by contouring the local osculating paraboloids. Away from any transition points, the intersection curves do not change topology and evolve according to a simple evolution vector field that is constructed in the euclidean space in which the surfaces are embedded. [ABSTRACT FROM AUTHOR]

The appeal of computer games may be enhanced by vision-based user inputs. The high speed and low cost requirements for near-term, mass-market game applications make system design challenging. In this paper we propose a vision based 3D racing car game controlling method, which analyzes two fists positions of the player in video stream from the camera to get the direction commands of the racing car. This paper especially focuses on the robust and real-time Bayesian network (BN) based multi-cue fusion fist tracking method. Firstly, a new strategy, which employs the latest work in face recognition, is used to create accurate color model of the fist automatically. Secondly, color cue and motion cue are used to generate the possible position of the fist. Then, the posterior probability of each possible position is evaluated by BN, which fuses color cue and appearance cue. Finally, the fist position is approximated by the hypothesis that maximizes a posterior. Based on the proposed control system, a racing car game, "Simulation Drive", has been developed by our group. Through the game an entirely new experience can be obtained by the player. [ABSTRACT FROM AUTHOR]

This paper describes the development of a real-time perceptive user interface. Two cameras are used to detect a user's head, eyes, hand, fingers and gestures. These cues are interpreted to control a user interface on a large screen. The result is a fully functional integrated system that processes roughly 7.5 frames per second on a Pentium IV system. The calibration of this setup is carried out through a few simple and intuitive routines, making the system adaptive and accessible to non-expert users. The minimal hardware requirements are two web-cams and a computer. The paper will describe how the user is observed (head, eye, hand and finger detection, gesture recognition), the 3D geometry involved, and the calibration steps necessary to set up the system. [ABSTRACT FROM AUTHOR]

In this paper we construct a novel human body model using convolution surface with articulated kinematic skeleton. The human body's pose and shape in a monocular image can be estimated from convolution curve through nonlinear optimization. The contribution of the paper is in three folds: Firstly, human model based convolution surface with articulated skeletons is presented and its shape is deformable when changing polynomial parameters and radius parameters. Secondly, we give convolution surface and curve correspondence theorem under weak perspective projection, which provide a bridge between the 3D pose and 2D contour. Thirdly, we model the human body's silhouette with convolution curve in order to estimate joint's parameters from monocular images. Evalution of the method is performed on a sequence of video frames about a walking man. [ABSTRACT FROM AUTHOR]

Image segmentation is an important research topic in medical image analysis area. In this paper, we firstly propose a generalized level set formulation of the Mumford-Shah functional by a sound mathematical definition of line integral. The variational flow is implemented in level set framework and thus implicit and intrinsic. By embedding a weighted length term to the original Mumford-Shah functional, the paper presents a generic framework that integrates region, gradient and shape information of an image into the segmentation process naturally. The region force provides a global criterion and increases the speed of convergence, the gradient information allows for a better spatial localization while the shape prior makes the model especially useful to recover objects of interest whose shape can be learned through statistical analysis. The shape prior is represented by the zero-level set of signed distance maps of images and is well consistent with level set based variational framework. Experiments on 2-D synthetic and real images validate this novel method. [ABSTRACT FROM AUTHOR]

In this paper, we present an easy method for automatic 3D garment pre-positioning. Given a digital 3D mannequin and a set of 2D patterns, the problem consists in sewing automatically the patterns around the mannequin and give an initial position of the garment that the mannequin has to wear. Our method is based upon the use of a figurine. A figurine of a virtual garment, which is always present in the garment design, is associated to the 2D patterns used for producing the garment. This figurine is a 2D drawing which gives the flavor and the style of the future garment. The figurine is the key element of the technique presented in this paper. It can be stored in a 2D CAD file with the 2D patterns of garments or in a database. In this method, we use two figurines, corresponding to the front and the back view of the future garment and a standard silhouette. The designer just needs to work once on the garment construction to allow the automatic pre-positioning. Once the designer has specified the links between the 2D garments and the figurine, he has just to take some simple actions on the figurine to complete the pre-positioning. Our method provides a mapping algorithm from the standard silhouette to a silhouette of a future client, which is the starting point of the pre-positioning. When finalized the automatic pre-positioning will help the designer to see directly the garment he invented on a "real" body. [ABSTRACT FROM AUTHOR]

In this paper, we present an experimental work using a MOCAP system and an iterative minimization technique to compute damping parameters and to measure their contribution for the simulation of cloth in free fall movement. Energy damping is an important phenomenon to consider for the 3D simulation of warp and weft materials, since it has a great influence on the animation realism. This phenomenon can be generated either by friction between moving cloth and air, or by friction between the warp and the weft threads of the fabric. The contribution of this paper is to determine viscous parameters of cloth using precise trajectory data of a real cloth. [ABSTRACT FROM AUTHOR]

In this paper, a new watermarking algorithm using 2D barcode is proposed. The 2D barcode means 2-dimensional barcode that contains more information than conventional 1-dimensional barcodes. For error correction, 2D barcode allow the recognition of barcodes that are up 60% damaged. Therefore, the 2D barcode as watermark can be survived when it used in highly noisy environment. In this paper, a new wavelet based watermarking algorithm using 2D barcode as watermark is proposed. Dynamically generated 2D barcode is inserted as watermark in wavelet domain. From the experimental result, the proposed algorithm shows better invisibility and robustness comparing with the conventional methods using plain image as watermark. [ABSTRACT FROM AUTHOR]

The paper presents theory and practical aspects of the detectors of characteristic objects in multi-channel images. It is based on the scale-space version of the structural tensor, adapted to operate on multi-channel signals. The method allows for object detection in N(2D signal space with additional respect to the scale-space. Responses of the structural tensor are composed in a linear weighted sum that allows for better signal discrimination. In such a unified tensor framework different feature detectors were defined for detection of lines, corners, lines in the Hough space, structural places, etc. Although the presented method was developed for road sing recognition is can be also used for detection of other regular shapes. The sought objects are defined by a syntactical description of building line segments and their connection type. The paper presents also experimental results and implementation details. [ABSTRACT FROM AUTHOR]

This paper presents a novel two-pass algorithm constituted by Linear Hashtable Motion Estimation Algorithm (LHMEA) and Hexagonal Search (HEXBS) for block base motion compensation. On the basis of research from previous algorithms, especially an on-the-edge motion estimation algorithm called hexagonal search (HEXBS), we propose the LHMEA and the Two-Pass Algorithm (TPA). We introduced hashtable into video compression. In this paper we employ LHMEA for the first-pass search in all the Macroblocks (MB) in the picture. Motion Vectors (MV) are then generated from the first-pass and are used as predictors for second-pass HEXBS motion estimation, which only searches a small number of MBs. The evaluation of the algorithm considers the three important metrics being time, compression rate and PSNR. The performance of the algorithm is evaluated by using standard video sequences and the results are compared to current algorithms. Experimental results show that the proposed algorithm can offer the same compression rate as the Full Search. LHMEA with TPA has significant improvement on HEXBS and shows a direction for improving other fast motion estimation algorithms, for example Diamond Search. [ABSTRACT FROM AUTHOR]

This paper presents a novel approach for automatic person labelling in video sequences using costumes. The person recognition is carried out by extracting the costumes of all the persons who appear in the video. Then, their reappearance in subsequent frames is performed by searching the reappearance of their costume. Our contribution in this paper is a new approach for costume detection, without face detection, that allows the localization of costumes even if persons are not facing the camera. Actually face detection is also used because it presents a very accurate heuristic for costume detection, but in addition in each shot mean shift costume localization is carried out with the most relevant costume when face detection fails. Results are presented with TV broadcasts. [ABSTRACT FROM AUTHOR]

The goal of the works described in this paper is to improve results produced by an object detector operating independently on each frame of a video document in order to generate a more robust index. Results of the object detector are "smoothed" along the time dimension using a temporal window. For a given frame, we count the number of occurrences of each object in the previous and next frames, and then only the objects whose number of appearance is above a threshold are validated. In this paper, we present a probabilistic approach for theoretically computing these thresholds. This approach is well suited to limit the number of false alarms provided by the static detector, and its principle of detection generalization also allows some detections that can be missed by the detector. [ABSTRACT FROM AUTHOR]

In this paper investigate how to preprocess method from input images for robust face recognition varying illumination environments. By training the different classifiers with different clusters of training data and adopting fusion method considering fitness correlation between clusters we found out better recognition performance than combining classifiers fed with same data. The proposed method tries to provide adaptive preprocessing as well as by exploring the filter selection and fusion based on illumination cluster. Illuminant based clustering is enhanced face recognition ratio. Face image is clustered several cluster unsupervised or statistical method and we adopt adaptive filter each cluster. In this paper, some cluster is preprocessed by single filter others and some cluster adopted preprocessing by filter fusion. We found that the performance of individual filtering methods for image enhancement is highly depending upon face image cluster. Also, in this paper we present the recognition system using the table of fitness correlations between clusters for combining the results from the individual clusters. We present examples from real applications for bad illuminant face images. [ABSTRACT FROM AUTHOR]

Image segmentation is discussed for years in numerous papers, but assessing its quality is mainly dealt with in recent works. Quality assessment is a primary concern for anyone working towards better segmentation tools. It both helps to objectively improve segmentation techniques and to compare performances with respect to other similar algorithms. In this paper we use a statistical framework to propose statistical measures capable to describe the performances of a segmentation scheme. All the measures rely on a ground-truth segmentation map that is supposed to be known and that serves as a reference when qualifying the results of any segmentation tool. We derive the analytical expression of several transition probabilities and show how to calculate them. An important conclusion from our study, often overlooked, is that performances can be content dependent, which means that one should adapt a measure to the content of an image. [ABSTRACT FROM AUTHOR]