最新刊期
卷 20 ， 期 4 ， 2015
- 摘要：Urban transportation has become a global challenge that plagues all global metropolises. Visual analytics techniques have recently become an important intelligent transportation technology that is crucial to the analysis and utilization of big transportation data. This study attempts to review the state-of-the-art of visual analytics techniques in urban transportation data comprehensively. These techniques have been developed since the proliferation of information visualization and visual analytics. The discussion mainly focuses on two broad aspects: road traffic analysis problems and other intelligent transportation-related problems. A detailed presentation of the visualization techniques and visual analytics systems is organized according to the transportation data type and transportation problem category. Recent research trends are succinctly summarized. Many early studies have focused on designing techniques,such as arrow graph, mosaic map, and traffic wall,to visualize road traffic. With these techniques, current research on road traffic analysis has placed emphasis on traffic events. However, the definition of traffic events remains limited to traffic congestion. Other application areas of visual analytics in the intelligent transportation domain include public transportation, traffic accidents, and human mobility. In recent years, a new research trend of mining and utilizing the social dimensions or social contextual information of vehicle trajectories or traffic events has emerged. From traffic flow visualization to visual analysis traffic incidents, from analyzing road traffic status to other urban transportation-related social problems, from analyzing single-source transportation data to multi-source data that are rich in social semantics, and from the traditional interactive visualization paradigm at the PC end to novel media and devices for visual presentation, the depth and breadth of research on transportation data visualization have significantly broadened. The research trend of this domain is also evident.关键词：urban transportation problems;visual analytics;GPS trajectory data;traffic flow analysis;traffic incident analysis5363|486|4更新时间：2024-05-07
- 摘要：Many issues still exist in data cleaning despite extensive studies on this method. With visual interface and visualization, visual data cleaning has become one of the most important data cleaning methods. This study describes existing data quality problems and visual data cleaning processes, reviews state-of-the-art visual data cleaning methods (including sources, categories of data quality issues,and visual data cleaning methods), and summarizes the challenges and opportunities associated with visual data cleaning problems. Data cleaning techniques are related to specific data quality issues. Thus,this study examines different data quality problems to summarize and review previous works on visual data cleaning. Based on data quality issues, visual cleaning methods are summarized as direct visual cleaning, visual missing data, visual uncertainty data, visual data transformation, and data cleaning resource sharing. Challenges and further research directions are surveyed according to different data quality issues. We introduce and provide an overview of visual data cleaning problems, as well as highlight research directions of visual data cleaning.关键词：data cleaning;visual cleaning;visual analysis;information visualization;data analysis3883|428|3更新时间：2024-05-07
- 摘要：The ray casting method is an important direct volume rendering method, but it depends on complex transfer functions. This paper presents a structural feature-based adaptive ray casting algorithm that can show inner structural features with simple transfer functions. We analyze scalar value along a ray to extract structural features, including feature segments.Subsequently,several heuristic rules (scale, order, and importance of feature segment) are used to determine the visibility of the feature segments.Furthermore, the opacity ofsample points iscalculated.Finally, the calculated opacity isused to perform volume rendering. Synthesis data, medical data, and industry CT are used in our experiments.Results show that our method is superior to other similar methods for structure exhibition, specifically for tiny structural features. Furthermore, our method allows the user to adjust several intuitive parameters. This possibility adds flexibility to our method. The presented algorithm is relatively slower than DVR but is fast enough for interactive rendering. The proposed structure-based adaptive ray casting algorithm allows users to reveal the inner features of volume data with a simple transfer function and intuitive parameters. Therefore, the proposed method can further improve the intuitiveness of the ray casting algorithm.关键词：volume rendering;structure feature;adaptive;ray casting3160|279|2更新时间：2024-05-07
- 摘要：Most existing medical systems provided by major commercial companies and open source communities depend on different operating systems and platform-related plug-ins.Thus, cross-platform access is difficult to provide. This study presents a browser-oriented medical image visualization system based on the latest web technology and HTML5. Our approach is designed on the basis of B/S mode. The proposed method employs a self-defined protocol to offer customized visualization services. In particular, we propose the Canvas technique for HTML4 and WebGL to accelerate browser visualization. We propose an asynchronous approach to provide progressive visualization. This approach constructs multi-resolution sampling data for the underlying dataset and employs an adaptive visualization scheme during user interactions. We tested our system using multiple clinical and medical datasets in different browsers. Results show that our system supports multiple browsers. Experiments on 2D and 3D visualization features show that our system can display 2D images in real-time (25 frame/s), as well as visualize 3D images interactively. For a dataset with a resolution of 512×512×154, low-resolution sampling performance achieves 60 frame/s), whereas that of high-resolution sampling is 1 frame/s). The proposed system fully supports cross-platform operation and is compatible with all browsers that support HTML5. These features significantly enhance user experience and openup prospects for remote and mobile medical image visualization systems, as well as give rise to a new opportunity for web medical image visualization systems.关键词：medical image;visualization;plug-ins;HTML5;WebWorkers;WebGL3931|286|1更新时间：2024-05-07
- 摘要：Large-scale multi-projector display systems offer high resolution, high brightness, a large field of view, and a compelling sense of presence. Thus, these systems have been considered effective choices to tackle the conflict between the increasing demands for super-resolution display and the resolution limitations of single display equipment. However, the display color and brightness of these systems show significant spatial variation, which can be very distracting, there by breaking the illusion of a single display. Furthermore, the use of the rear projection mode has serious an isotropy problems; in particular, the brightness of the projected image varies in accordance with the location of the observer. In this paper, we propose an effective solution to solve the an isotropy problem of rear-projection multi-projector tiled display walls. Our solution includes two levels:static and dynamic optimal observation districts. The static optimal observation district method is focused on situations with a small amount of audience who only moves in a small area. Geometry and color calibrations are executed in accordance with a predetermined observation point by utilizing an information processing model of human attention. Subsequently,observers in the neighborhood of the observation point can see calibrated images with highly visual seamless effects.To relax the restrictions of the moving range, we present an observer-centered framework for the dynamic optimal observation district by integrating human tracking techniques and fuzzy predictive control algorithms. In this framework, we first optimize the process of creating projector models and color look-up tables. The cente positions of the target observer are then computed viavideo-based object tracking. Finally, fuzzy predictive control algorithms are used to deal with the positions to create step-shaped coordinates for the human body center,there by minimizing the need to adjust the brightness compensation and to speed up the screen calibration. The conventional color calibration methods only work well for certain special positions before the display surface because the color calibration results, which are usually organized as a color look-up table, are obtained off line and the audience's positions are not considered. Extending these methods, we introduce the static optimal observation district method, which allows users to preset an observation point. Furthermore, we allow the user to set the observation point online by capturing the observer's positionsin real-time.To test our methods, a screen calibration system for a rear-projection multi-projector tiled display wall is presented on the basis of the dynamic optimal observation district. Experimental results show that calibration methods with a dynamic optimal observation district have good robustness and high calibration efficiency.Therefore,our algorithm generally obtains better calibration results than other similar methods. Rear-projection multi-projector tiled display walls can be used in several applications. Anisotropy during color calibration has always been a challenge to researchers. This paper introduces the real-time feed back mechanism and proposes a novel dynamic strategy for solving anisotropy problems. The proposed solution can reduce the adverse effects of anisotropy on screen calibration and greatly improve the output image. The experimental results indicate that our solution has a preferable calibration effect even insituations with large-scale movements and has acceptable real-time calculations.关键词：multi-projector tiled display wall;rear projection;anisotropy;optimal observation point2794|255|0更新时间：2024-05-07
- 摘要：A fusion rule must be reasonably designed to make full use of complementary information, namely, the rich texture information in visible images and the strong object-targeting information in infrared images. A visible and infrared image region-level feedback fusion algorithm is proposed on the basis of performance evaluation.The images to be fused are initially decomposed into their low-frequency and high-frequency parts via the nonsubsampled contourlet transform(NSCT). Meanwhile, fractal features are adopted to perform man-made object enhancement of infrared images.Object and background areas are then obtained through threshold segmentation. In the design of low-frequency fusion rule, the weighted fusion coefficients of the object and background areas are selected as parameters.A genetic algorithm is used to optimize these parameters in accordance with the performance evaluation of fusion results. A regional weighted average fusion rule is used to fuse high-frequency parts. Finally, inverse NSCT is performed to obtain the fusion image with the optimized parameters. Three sets of images are used to compare the performance of four fusion algorithms by subjective and objective evaluation. Experimental results demonstrate that the fusion images from the proposed algorithm are natural and that the objects in the images are notable.Furthermore, the objective evaluation result is optimal. The proposed algorithm can combine the object information from infrared images and the background information from visible images.The resulting fusion images have a strong contrast, which is beneficial for battlefield situation display and object recognition tasks.关键词：image fusion;performance evaluation;nonsubsampled contourlet transform;fractal feature;genetic algorithm3149|448|1更新时间：2024-05-07
- 摘要：Images captured in foggy weather are of ten degraded byatmospheric absorption and scattering.Haze removal is highly desired in image processing and computer vision applications. Removing haze can significantly increase the visibility of the scene. In addition, most image processing and computer vision algorithms usually assume that the input image is the scene radiance. Therefore, several methods for haze removal have been proposed.However, the sky region processed by most of these algorithms is degraded by block noise and serious color distortion. To address this issue, this paper proposes an improved single image haze removal method based on sky region detection and dark channel prior. Our proposed method consists of three major stages: sky region detection, haze removal, and tone mapping. In the first stage, sky is usually a large and smooth region with high intensity. On the basis of these characteristics, an effective algorithm is designed to divide the input image into “sky” and “non-sky”regions. In the next stage, dark channel prior is used to estimate the transmission maps of the two regions, and a guided filter is applied to refine these maps, such that the haze-free image can be recovered by the atmosphere scattering model. The final stage uses a simple tone mapping algorithm to increase the image brightness, which leads to good visual effects. In He's haze removal algorithm, dark channel prior is no longer a good prior because sky regions may have high intensity.Consequently,the sky region of the recovered haze-free image will have serious noise and color distortion. We combine sky region detection and dark channel prior to eliminate noise and color distortion. Several experiments show that images restored by the proposed algorithm are clear and natural. In particular, the sky region is smooth and bright. Dark channel prior is a very good prior for image haze removal, but it is not suitable for sky regions because it leads to block noise and serious color distortion. On the basis of sky region detection, a novel single image haze removal algorithm is proposed in this paper. The presented algorithm can achieve better results than the defogging algorithms proposed by He Kaiming and Tarel.关键词：sky regions detection;dark channel prior;haze removal;tone mapping5684|701|25更新时间：2024-05-07
- 摘要：Human action recognition is an important research topic in the field of computer vision;this method has promising potential applications. On the basis of the limitations of local and global spatial-temporal features, a novel and effective middle-level spatial-temporal feature is proposed for action recognition. The proposed feature encodes the structural distribution of local features in the neighborhood of the spatial-temporal interest point (STIP), there by improving the discriminative power of STIP.This feature can model the flexible intra-action variations. Pointwise mutual information is introduced to measure the correlation between the mid-level feature and the action.The video clip is finally classified as the action category that has the greatest mutual information with the mid-level features. Experimental results validated the advantage of the proposed mid-level feature over the local-feature-based baseline methods and other published results. The mid-level feature achieved 96.3% and 98.0% recognition accuracies on the KTH and ADL(Activities of daily living) action datasets, respectively. The proposed mid-level spatial-temporal feature enhances the discriminative power of actions by harnessing the spatial-temporal distribution of local spatial-temporal features.Consequently, this feature is capable of recognizing realistic human actions.关键词：action recognition;spatial-temporal interest point;mid-level spatial-temporal feature;pointwise mutual information3588|908|0更新时间：2024-05-07
- 摘要：In GPUs based on TBR architecture, the efficiency of triangle raster greatly influences chip performance. Traditional algorithms generate severalredundant pixels and lose the advantages of TBR architecture. A high-efficiency triangle raster algorithm is proposed in this paper. The features of TBR architecture are fully utilized in this algorithm. The draw parameters of each tile are calculated in the pre-processing stage. The position relationship between the triangle and the tile boundary is obtained. All data are committed to memory. In the raster stage, the Bresenham algorithm is used to obtain all the horizontal scan lines in every tile untilall pixels of the scan lines are eventually generated. After theoretical analysis, the raster efficiency of the raster operation can reach 83% and even approach 100%,depending on the triangle shape. The function and performance of this algorithm are evaluated with a FPGA proto type verification system. To adapt to the TBR architecture, a new triangle raster algorithm is proposed in this paper. The pixel fill rate is equivalent to ATI M9, which has twice the clock rate. Therefore, the algorithm can achieve high raster efficiency.关键词：triangle;raster;tile based rendering;prototype verification4264|304|2更新时间：2024-05-07
- 摘要：On the basis of a new orthogonal function system called V-systemover a triangular domain, this study proposes a frequency-based watermark algorithm for 3D models. Vertex information is acquired to construct “pseudo model” is constructed and is then parameterized so that the pseudo model can be expressed as a piecewise linear function over a triangular domain. The linear function is then orthogonally decomposed into a V-series, which has coefficients from which the V-spectrum is obtained. Finally, the watermark is embedded into the frequency domain by modifying the V-spectrum, such that a renewed V-spectrum is formed. The pseudo model is reconstructed by using the new V-spectrum and the basis function of the V-system, and the final watermarked model is built by replacing the vertex of the original 3D model by the vertex information of the watermarked pseudo model. The watermarked model created by the proposed algorithm has high fidelity, which is maintained even when multiple watermarks are embedded into a model.Moreover, the bit error rate of the extracted watermark is low, and the correlation coefficient between the original watermark and the extracted watermark is high. The proposed algorithm is theoretically proven to be robust against similarity transformations. Experiments conducted on the watermarked model show that the proposed method is robust against noise pollution, shearing, and smoothing.关键词：3D model;digital watermark;V-system;orthogonal decomposition;spectrum3985|560|1更新时间：2024-05-07
- 摘要：Obtaining an image that contains all relevant objects in focus is often made impossible by the limited depth of field of macro lens. Consequently, the obtained image will fail to be in focus for all objects, i.e., if one object in the scene is in focus, another object will be out of focus. Macro images should be captured and fused with different degrees of focus to obtain a clear image for macro photography. Most multi-focus fusion algorithms assume that source images possess point-wise correspondence, i.e., the colors at and around any given pixel in one image correspond to the colors at and around that same pixel in another image. However, when a mechanical device is used to capture different in-focus images, small motion inevitably occurs between adjacent images.This study proposes a multi-focus image capture and fusion system for macro photography. The system consists of three parts. The first component is a multi-focus image capture device that can capture a series of macro images with high precision. These images are taken at different focus distances from a photographic subject. The second component is a feature-based method that can automatically align multiple in-focus images. The third component is a multi-focus image fusion method that combines multiple images taken and aligned previously with a fused image with a large depth of field. The proposed fusion method is based on Gaussian and Laplacian pyramids with a novel weight map computation strategy.A multifocus image fusion method based on multiresolution can be obtained by combining weight calculation with canonical image pyramid. Several data sets are captured and tested with the use of the proposed system to verify the soundness of hardware and software design. Subjective and objective methods are also used to evaluate the proposed system. According to the subjective evaluation, the fused macro image generated by the system not only has sufficient depth of field but can also clearly present small details of the object at high resolution. According to the objective evaluation, the synthesized macro image of the system is optimal in all three types of evaluation criteria, namely,standard deviation, average gradient and information entropy when compared with those obtained with other methods. An analysis of the experimental results shows that this system is flexible and efficient.The system can acquire, register,and fuse multiple multi-focus macro photos with fused image quality that is comparable with that of other methods.关键词：macro photography;multi-focus image;image alignment;image fusion4749|273|1更新时间：2024-05-07
- 摘要：Shadows usually degrade image quality and cause undesirable problems. Hence, shadow detection is a fundamental step in computer vision and image analysis, including such processes as image segmentation, object recognition, stereo registration, and scene analysis. For a single image, shadow detection is particularly challenging because of limited information. Most shadow detection algorithms have difficulty in detecting lathy shadows and self-shadows,as well as in distinguishing between shadows and dark pixels. To address these problems, a novel algorithm with pairwise regions for shadow detection is proposed in this study. Unlike traditional algorithms that explore pixel or edge information, the proposed algorithm involves the training of two models with support vector machine to learn shadow features and to classify shadow and nonshadow regions. Our algorithm has two stages: offline learning and online detecting. In the offline stage, the image is first segmented,after which every single regionis obtained by using the mean shift and canny detection algorithms. Support vector machine is then employed to construct a single region shadow model with the use of the texture and intensity features in each region. A pairwise region shadow model is finally constructed after manually marking pairwise regions of shadow and nonshadow with the distances of texture histograms, color ratios(in RGB color space and lab color space), and the ratios of H channel to I channel in HSI color space. In the online stage, the same segmentation manipulation as that in the prior stage is performed for the input image. Thereafter, the features of the single and pairwise region models are extracted and integrated into the corresponding model to obtain the classification results separately. Finally, a graph is built using the two models, and the graph-cut algorithm is employed to label shadow and nonshadow regions. The following are the advantages of our method: 1) We consider both pixels and edges to achieve accurate segmentation, particularly for long and thin shadows; and 2) Except for common shadow features, we employ the ratios of H channel to I channel in HSI color space to detect self-shadows and to remove dark pixels from shadows. Visual experimental results show that our algorithm not only detects spindly shadows and self-shadows effectively but also separates shadows from dark pixels correctly. In terms of confusion matrix in shadow detection, our algorithm achieves an 85.2% performance versus 70.1% for the algorithm reported by Guo et al. and 60.2% for the algorithm by Tian et al. In addition, our algorithm runs 34% faster than that of Guo et al. under the same situation because of the use of a simple feature set. Shadow detection algorithms based on regions are commonly used for outdoor image processing. However, few algorithms can detect some special shadows, such as threadlike shadows and self-shadows, or to distinguish shadows from dark pixels. In this study, a new algorithm is presented to solve the problems arising from a single outdoor image by using paired regions. Experiment results indicate that our algorithm has satisfactory performance in detecting spindly shadows and self-shadows, as well as in distinguishing shadows from dark pixels.关键词：shadow detection;shadow model;shadow feature;image segment5083|541|5更新时间：2024-05-07
- 摘要：Most materials have some types of fine structures, which inevitably increase the complexity of realistic rendering. In this study, these complex structures are further categorized into three scales, namely,macroscale, mesoscale, and microscale. Each scale is modeled with different strategies. In particular, the macrostructures are modeled with traditional triangle meshes because these structures are large and visible under most circumstances. The mesostructures, which are tiny but still resolvable in the image,are represented by normal maps, whereas the smaller microstructures are directly depicted with a single roughness parameter because these structures have a significantly smaller size than the pixel width. For each scale, the normal distribution function (NDF) is obtained and fitted with a vMF distribution or mixtures of vMF distributions. Thereafter, the final NDF is approximated by convoluting each NDF. Furthermore, the paraboloid map and summed-area table are adopted to support real-time rendering of dynamic scenes, including time-varying lighting and deformed meshes. Experimental results reveal that the proposed approach can generate photorealistic surface reflections under varying viewpoints at a real-time frame rate. A real-time rendering method for rough surfaces with multi-scale properties is proposed in this study. Our method supports dynamic lighting and deformable objects.关键词：multi-scale;rough surface;dynamic lighting;real-time rendering;filtering4546|341|0更新时间：2024-05-07
- 摘要：The volume parametric model has many excellent characteristics and extensive application prospects. Determining the method by which to create the model is an urgent issue to address. This study first presents the expression of volume parametric model and proposes the setting of control points as the core issue. After the control points of the six boundary surfaces are presented, a new method for creating the control points of the volume parametric model is presented on the basis of the differential form of harmonic functions from the computational domain to the parameter domain. The Jacobian matrix and isoparametric network are also presented to assess the quality of the generated volume model. A volume parametric model is constructed on the basis of the point cloud model.A detailed comparison is performed between our method and the discreet Coons and convex combinative interpolation methods. Results show that the discreet harmonic function method can obtain a more stable and optimized model compared with other methods. Discrete harmonic mapping method can get initial volume parametric model but the model is required to be optimized and has provides a very good basis by using our method.关键词：volume parameterization;control point setting;harmonic mapping;mesh quality4919|317|3更新时间：2024-05-07
- 摘要：Matching and extracting the target curve from a large dataset are time-consuming and laborious tasks when traditional feature detection and tracking methods based on the temporal curve analysis are used because prior knowledge is required to determine the shape of the temporal curve of the region of interest.In this study, we introduce a novel method based on asymmetric Gaussian function to conduct real-time feature detection and tracking for time-varying volume visual analysis. A low-pass filter is first employed to smoothen the original temporal curve, which helps achieve the minimum data points accurately. The original temporal curve can be further subdivided into several segments according to the minimum points. We then take advantage of asymmetric Gaussian functions to fit the segments and to achieve a new Gaussian temporal curve, which can be further expressed on the basis of several simple Gaussian parameters. A convenient interface in which users can select the features of interest by means of visibility analysis along the viewing rayis also provided. Therefore, the evolution of the selected features can be further tracked in different time steps by matching the corresponding Gaussian parameters. To demonstrate the effectiveness of the proposed method, time-varying simulative data are used to conduct feature detection and interactive visualization. A large number of experimental results indicate that the proposed method can detect time-varying features accurately and quickly. The proposed technique expresses the basic temporal curve with new asymmetric Gaussian temporal curve and then takes the simple Gaussian parameter matching as a substitute for the complicated temporal curve matching process. Compared with traditional methods, the proposed approach can conduct feature detection and tracking in real time.关键词：time-varying volume visualization;temporal curve;feature tracking;asymmetric Gaussian function5634|282|1更新时间：2024-05-07
- Postal code：100190
- Tel：010-58887035/58887030/58887418 Email：jig@aircas.ac.cn
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备05080539号-4
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0