Bing-Yu Chen | National Taiwan University (original) (raw)

Papers by Bing-Yu Chen

Lecture Notes in Computer Science

Digital photos are massively produced while digital cameras are becoming popular, however, not ev... more Digital photos are massively produced while digital cameras are becoming popular, however, not every photo has good quality. Blur is one of the conventional image quality degradation which is caused by various factors. In this paper, we propose a scheme to detect blurred images and classify them into several different categories. The blur detector uses support vector machines to estimate the blur extent of an image. The blurred images are further classified into either locally or globally blurred images. For globally blurred images, we estimate their point spread functions and classify them into camera shake or out of focus images. For locally blurred images, we find the blurred regions using a segmentation method, and the point spread function estimation on the blurred region can sort out the images with depth of field or moving object. The blur detection and classification processes are fully automatic and can help users to filter out blurred images before importing the photos into their digital photo albums.

ACM SIGGRAPH 2004 Posters on - SIGGRAPH '04, 2004

ACM SIGGRAPH 2006 Research posters on - SIGGRAPH '06, 2006

In this paper, we propose a novel graph representation, Railway Visualization, to reveal the info... more In this paper, we propose a novel graph representation, Railway Visualization, to reveal the information evolution with temporal relationship over time. Through Railway Visualization, it cannot only reveal the changes of an entity's intra-state over time, but also can effectively disclose the intrinsic characteristics of the complicated relationship evolution, such as branching, crossing, splitting, converging and parallelism. Using the railway route map in people's daily life as the metaphor to visualize the information evolution can leverage people's existing knowledge and mental models to quickly interpret the graph and gain the insight. We then conclude the drawing aesthetic rules by observing the railway route maps and formulate them as the energy terms. By solving them through a two-steps optimization method, the railway-route-map-style layout can be rendered automatically. Furthermore, an interactive level-of-detail mechanism is also provided. Finally, several types of data, including movies, photo albums, news, and co-author network are visualized to prove the virtues of the Railway Visualization.

2013 International Conference on Cyberworlds, 2013

Figure 1. Interaction with the Thememory. (a) The synthesized program theme by utilizing a user's... more Figure 1. Interaction with the Thememory. (a) The synthesized program theme by utilizing a user's photos. (b) A photo is hovered. (c) Some photos related to the selected one are popped up.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2009

In this paper, a semi-automatic method to capture motion data from a single-camera video is propo... more In this paper, a semi-automatic method to capture motion data from a single-camera video is proposed. The input video is first segmented and analyzed, and a 3D character model with skeleton rigged is used as a reference model. Then, the reference model is modified to fit the subject's contour in the starting frame, and the body's and limbs' contours of the subject are also specified by the user. Our system then extracts the motion from the video by estimating the reference model's poses automatically in each video frame forwardly. Finally, the user can help to refine the result through a friendly user interface.

Conference on Human Factors in Computing Systems - Proceedings, 2010

This work describes a novel approach to utilizing everyday objects of users as additional, auxili... more This work describes a novel approach to utilizing everyday objects of users as additional, auxiliary, and instant tabletop controllers. Based on this approach, a prototype platform, called iCon, is developed to explore the possible design. Field studies and user studies reveal that utilizing everyday objects such as auxiliary input devices might be appropriate under a multi-task scenario. User studies further demonstrate that daily objects can generally be applied in low precision circumstances, low engagement with selected objects, and medium-to-high frequency of use. The proposed approach allows users to interact with computers while not altering their original work environments.

SIGGRAPH Asia 2012 Technical Briefs, SA 2012, 2012

One of the holy grails of computer graphics is the generation of photorealistic images with motio... more One of the holy grails of computer graphics is the generation of photorealistic images with motion data. To regenerate convincing human animations might not be the most challenging part, but it is definitely one of ultimate goals for computer graphics. Amongst full-body human animations, facial animation is the challenging part because of its subtlety and familarity to human beings. In this paper, we like to share the work of lip-sync animation, part of facial animations, as a framework for synthesizing lip-sync character speech animation in real time from a given speech sequence and its corresponding texts.

Ninth International Conference on Computer Aided Design and Computer Graphics (CAD-CG'05)

In this paper, we propose a new framework for developing 3D graphic applications on portable devi... more In this paper, we propose a new framework for developing 3D graphic applications on portable devices such as cellular phones or personal digital assistants (PDAs). Our framework is based on J2ME (Java 2 Platform, Micro Edition 1) environment, an application environment that specifically addresses the needs of developing programs on embedded systems using Java, and we leverage the advantages of J2ME to build our platform-independent 3D graphic library on mobile devices. Adopting this framework, the 3D program developers can not only implement a cross-platform 3D program easily as they did using OpenGL ES 2 library but also gain the powerful features including efficient memory usage and shading functionalities.

2006 Digest of Technical Papers International Conference on Consumer Electronics

To provide 3D programs for all kinds of mobile devices, platform-dependent is still a big problem... more To provide 3D programs for all kinds of mobile devices, platform-dependent is still a big problem. In this paper, we present a platform-independent 3D graphics library, jGL ES, which is based on J2ME and has a similar programming environment as OpenGL ES. With jGL ES, the programmers can follow the same programming style of the framework of OpenGL ES and smoothly transfer their source codes to be cross-platform applications.

2010 International Computer Symposium (ICS2010), 2010

In this paper, we provide a system that takes start and end points as the input, and will automat... more In this paper, we provide a system that takes start and end points as the input, and will automatically connect to the Google Maps with Street View to achieve the route planned result and scenery along the route. Finally, it will generate a smooth scenic video from the starting point to the destination, which combines the Google Maps to provide better route recognition to users. The users can watch and interact with the video as if they are driving a car through the planned route. Our system is fully automatic but still provides interaction mechanisms for the users to walk around in the scenic video.

IEEE Internet Computing, 2010

Computer Graphics Forum, 2012

There is considerable recent progress in hair simulations, driven by the high demands in computer... more There is considerable recent progress in hair simulations, driven by the high demands in computer animated movies. However, capturing the complex interactions between hair and water is still relatively in its infancy. Such interactions are best modeled as those between water and an anisotropic permeable medium as water can flow into and out of the hair volume biased in hair fiber direction. Modeling the interaction is further challenged when the hair is allowed to move. In this paper, we introduce a simulation model that reproduces interactions between water and hair as a dynamic anisotropic permeable material. We utilize an Eulerian approach for capturing the microscopic porosity of hair and handle the wetting effects using a Cartesian bounding grid. A Lagrangian approach is used to simulate every single hair strand including interactions with each other, yielding fine-detailed dynamic hair simulation. Our model and simulation generate many interesting effects of interactions between fine-detailed dynamic hair and water, i.e., water absorption and diffusion, cohesion of wet hair strands, water flow within the hair volume, water dripping from the wet hair strands and morphological shape transformations of wet hair.

Animating pictures is an interesting and useful visual effect in entertainment industry. In this ... more Animating pictures is an interesting and useful visual effect in entertainment industry. In this paper, we present a system that allow user to animate a character in pictures in 3D space by applying 3D motion data. We use a 3D character mesh with skeleton rigged as template model. The user need to cut out silhouette of the character in pictures, and assign correspondence points between 2D character image and 3D template model. System then fits the template model to the image. Finally, the user can apply any 3D motion data to create animations in 3D space.

This paper presents a semi-automatic method to capture animation data from a single camera video.... more This paper presents a semi-automatic method to capture animation data from a single camera video. The video is first segmented and analyzed, and a 3D character model with skeleton rigged is input as a reference model. Then, the user modifies the reference model to fit the subject's contour in the starting frame, and specifies the body and limbs contours of the subject character. Our system then reconstructs the motion by estimating the reference model's pose automatically in each frame forward. Finally, the user's intervention helps to refine the result. This paper shows a better motion reconstruction result than totally automatic methods that are widely used in video-based motion capture in computer vision.

Web Graphics has become an important and new platform today, since there are more and more people... more Web Graphics has become an important and new platform today, since there are more and more people that would like to have 3D graphics supports on the Web. In order to distribute 3D graphics applications from servers to clients on different platforms, a general-purpose 3D graphics library for Java called jGL is developed as the base of the 3D graphics applications for Web Graphics. To make jGL to be easy to learn and use, its API (Application Programming Interface) is designed in a manner quite similar to that of OpenGL. Therefore, the programmers who are familiar with the usage of OpenGL can use jGL intuitively since they can find one-to-one mapping functions in both of them. Besides the functions of OpenGL, jGL also supports other useful functionalities, such as Phong shading, bump mapping, environment mapping, procedural solid texturing, and VRML (Virtual Reality Modeling Language) as its extensions. First of all, I would like to deeply appreciate Prof. Tomoyuki Nishita, my advisor. It is my pleasure and so lucky to be his first PhD. student in the University of Tokyo. He taught me not only for how to do the researches, but also for how a good researcher and person should be. To be a good researcher as him is my next dream. Of course, I would also like to thank all of the members and past members in our laboratory, especially for Prof. Jieqing Feng, Yosuke Bando, and Yutaka Ono, my colleagues. We did really enjoy the co-research life so much. Prof. Yoshinori Dobashi usually comes to our laboratory and gives me a lot of suggestions; I would also like to thank him here. Then, I wish to express my best appreciations for Prof. Ming Ouhyoung, the advisor of my Master thesis, Prof. Ja-Ling Wu, and Prof. Wen-Chin Chen. They do really give me many suggestions and commends for researches and personalities even when I left the National Taiwan University. I was so lucky to have the chances to work with them and also several ex-members of the Communications and Multimedia Laboratory, especially the Graphics Group ex-members and assistants.

To date, more high resolution animated models are required to present important details and fine ... more To date, more high resolution animated models are required to present important details and fine structures, however, sometimes such high resolution models are unnecessary and undesired. For example, we usually want to preview a low resolution animated model to decide if we want to download it or not. Interactive systems, for another example, sometimes use low resolution models to obtain better performance. Though there are many wellknown algorithms dealing well on simplifying 3D models, most of them are limited to static ones. Applying these mesh simplification methods to 3D animated models, a good simplified model in a specified pose can be obtained. However, some features of the original animated model, which can be shown in other poses, may be destroyed. In this paper, we propose an automatic method to simplify a 3D animated model which takes the features shown in every poses into account and preserves the geometry details of it.

X3D (Extensible 3D) is a standard description language for describing 3D scenes on the web enviro... more X3D (Extensible 3D) is a standard description language for describing 3D scenes on the web environment and can be used to establish 3DVR e-commerce systems, such as virtual shopping malls. However, due to the CPU and memory limitations of mobile devices, the 3DVR e-commerce systems described by using X3D scene language cannot be used on mobile devices. This limitation delays the growth of 3DVR e-commerce systems, since the population and influence of mobile devices are more powerful than PC environments. To make X3D to be really cross platform, we provide a cross-platform X3D (CP-X3D) environment in this paper, which including a modified X3D scene language suitable for mobile devices and a cross-platform CP-X3D engine for both of mobile devices and PC environments. Through the CP-X3D environment, the 3DVR e-commerce systems described in CP-X3D scene language or developed with CP-X3D engine can be used on both of PCs and mobile devices.

Facial animation is traditionally considered as important but tedious work for most applications,... more Facial animation is traditionally considered as important but tedious work for most applications, because the muscles on face are complex and dynamically interacting. Although there are several methods proposed to ease the burden from artists to create animating faces, non of these are fast and efficient in storage. This paper introduces a framework for synthesizing lips-sync facial animation given a speech sequence. Starting from tracking features on training videos, the method first find representative keyshapes that is important for both image reconstruction and guiding the artists to create corresponding 3D models. The training video is then parameterized to weighting space, or cross-mapping, then the dynamic of features on the face is learned for each kind of phoneme. The propose system can synthesis lips-sync 3D facial animation in very short time, and requires very small amount storage to keep information of the key-shape models and phoneme dynamics.

In the first part of this paper, we describe a novel method of panoramic view rendering for real-... more In the first part of this paper, we describe a novel method of panoramic view rendering for real-time interactive applications trying to present a real-world like environment. The proposed method is different from the approach of QuickTime® VR, in that the scene is rendered by generating a sphere-like polyhedral environment map from photo-realistic images and using the generated maps to