NAIST Graduate School of Information Science
| Date: | Thursday, Aug. 25, 2005 |
|---|---|
| Time: | 13:30 - 16:00 |
| Place: | L1 Lecture Room |
| Language: | English (Oral Presentation), English/Japanese (Question) |
| Chairperson: | Kentaro Takemura (Robotics Lab. : PD), Thomas Clouqueur (Computer Design and Test Lab. : PD) |
In this presentation, I will present a self-controlling active camera system intended for a free viewpoint 3D video application. My concern is, given an active camera system made up of cameras with long focal length lenses (i.e. only partial views of the moving subject can be gotten), how to endow the said system with the possibility to change its configuration in order to 1) follow the subject's movement within the scene, and 2) to assign each camera to a specific part of the moving object so as to allow the best visibility of the whole subject. The aim is to release the resolution from the limits imposed by the use of fixed camera systems.
I will present a visibility-based camera assignment scheme. It consists in analyzing the visibility of the 3D shape obtained by a real time 3D reconstruction process at a given time, and accordingly, deducing the new parameters to which our camera system will be set to in the following step.
Recently, networked telepresence which allows multiple users in different places to see a virtualized real world in network environment has been investigated with high speed network and high performance PC. Conventional networked telepresence systems using omni-directional video streams captured by an omni-directional camera enable users to see a virtualized remote site in arbitrary direction. The systems use uni-cast protocol for video transmission or transmit each user's view from server, so network traffic and load of server increase when many users use it. In this presentation, I propose a new telepresence system which enables multiple users to look around a virtualized remote site interactively using web browsers and multi-cast of omni-directional video streams.
With the spread of digital cameras, shooting photos has been becoming an everyday affair. However, there are few methods or systems to manage photos simply, and a huge amount of photo data remains unorganized. Although it is possible to add appropriate words explaining the contents of the photo as one of the methods to manage photos, it requires much time and effort to input such captions manually. It is also difficult to add captions intended by a user automatically. In this presentation, we propose a semi-automatic photo captioning system that enables users to generate captions simply and browse a photo library efficiently. Caption candidates are acquired from map database retrieval and relevant word extraction using web retrieval based on shooting position and orientation information. We have implemented a captioning system prototype based on the proposed method, and have carried out some experiments.
In applications of wearable computers like a wearable augmented reality (AR) system, the exact position and orientation of a user are required for location-based information presentation. One of the methods estimates the user's position and orientation by recognizing the visual markers which are pasted up on the ceilings or walls. However, the method has problems concerning undesirable visual effects. In order to avoid the problems, we propose a localization method which is based on using an infrared camera and invisible visual markers consisting of translucent retro-reflectors. In the proposed method, user's position and orientation are identified by recognizing invisible visual markers which are illuminated from infrared LEDs.
We are researching a novel sound interface of man-machine communication, which realizes high-quality sound reproduction and noise-free sound capturing simultaneously. Our proposed method is a combination of sound field reproduction technique and microphone array signal processing. Since the sound field reproduction can reproduce any expected signal accurately at control points, virtual reality of sound can be presented to a user. In addition, by representing silent signal at the positions of microphones, the response sound is prevented to be recorded by the microphones. After the observed signals of microphones are processed to cancel the environmental noise, an output signal includes only the user's speech and its speech recognition performance is improved. The experimental result shows the efficacy of the proposed method.
The goal of this study is to analyze the relationship between the level of program comprehension and the bug detection rate in a code review process. We examined the reviewer's comprehension level of a program after the review process was held by the reviewer. As a result of examinations, we have found that the bug detection rate correlated with the comprehension level of data structure and its manipulation while it did not correlated with the comprehension level of logic and interface. We also found the program comprehension was a necessary condition for detecting bugs.
The goal of this presentation is to give a guideline of applying light-weight obfuscation methods to protect cipher programs against the cracker who tries to extract security-sensitive data. We focus on the crackers' viewpoint including crackers' knowledge, cracking tools and conjectures concerning the cipher programs. The point is that, this is a cracker-centric approach focusing on "disrupting" the cracker's actions, while conventional obfuscation methods are protector-centric approach focusing on building a "complex" program.