NAIST Graduate School of Information Science
| Date: | Friday, Feb. 24, 2006 |
|---|---|
| Time: | 13:30 - 16:20 |
| Place: | L1 Lecture Room |
| Language: | English (Oral Presentation), English/Japanese (Question) |
| Chairperson: | Mika Naganawa (Image Processing Lab.: PD) Kentaro Takemura (Robotics Lab.:D3) |
Acquiring image data with global position and orientation can be used for prior knowledge database for localization of mobile camera users in ubiquitous computing environments. One of main problem in conventional camera pose estimation methods is accumulative errors. This presentation reports experimental results of pose estimation method of an omnidirectional multi-camera system using feature tracking and GPS data in order to avoid the accumulative error problem. We will demonstrate the proposed method enables us to estimate camera pose information in the same order precision as a conventional method using landmarks.
This presentation shows a new view management method for annotation overlay using augmented reality(AR) systems and evaluating experiments. The proposed method emphasizes the user-viewed object and the corresponding annotation in order to present links between annotations and real objects clearly. This method includes two kinds of techniques for emphasizing the user-viewed object and the annotation. First, the proposed method highlights the object which is gazed at by the user using a 3D model without textures. Secondly, when the user-viewed object is occluded by other objects, the object is complemented by using an image made from a detailed 3D model with textures. This presentation also shows experiments which evaluate highlighting techniques of user-viewed objects.
The program(e.g., cipher module) which contains security-sensitive information is increasing. and then, the demand for protecting programs against the cracker who tries to extract the information is increasing, too. Program obfuscation is a one of techniques to protect and it is proposed variously and numerously. But most proposed methods are unclear, how effective to protect. In this presentation, We establish a goal-oriented analysis framework for proper use of the existing obfuscation techniques. First, we assume an imaginary cracker with his purpose and target (i.e., goal). Then, we break down the goal into pieces and we apply an appropriate obfuscation to each.
Abstract:The Japanese digital television terrestrial broadcasting (DTTB) based on orthogonal frequency division multiplexing (OFDM) is sensitive to Doppler spread and multi-path propagation in mobile environment. A Doppler spread compensator based on linear array antenna has been proposed. However, the proposed linear array antenna assisted Doppler spread compensator did not take into account an effect of electromagnetic coupling among array elements. Also, it did not have diversity function while it has a several array elements. To mitigate bit error rate (BER) performance degradation due to electromagnetic coupling effect, we propose a mutual coupling cancel method by making use of dummy elements add on both sides of monopole array. Besides, we perform diversity function to improve BER performance in multi-path fading environment. In this presentation, we present performance improvement methods of integrated services digital broadcasting for terrestrial (ISDB-T) in fast fading environment.
In this presentation, we propose a fast and low complexity algorithm of computation the combining weight of the receiver's combiner without divisions operations. This works is in the conjunction of our previous method by replacing Carrier Interferometry (CI) Spreading by an FFT spreading. However, in this presentation, we extend the works to the area of combiner at the receiver. Combiner is widely used at the receiver of Carrier Interferometry OFDM (CI/OFDM), Spread OFDM and Multi-carrier Code Division Multiple Access (MC-CDMA). Combiner is essential for recovering the transmit signals which is corrupted by the fading channel that has a multiplicative effect. For this purpose, combiner's weight should be obtained from the division or inversion of the channel impulse response to avoid the multiplicative effect in the faded channel. For the next generation of high bit rate applications, these divisions are necessary to be computed in a very short time and may impact to the increasing of hardware complexity. The main contribution of this paper is a proposed fast algorithm by replacing the divisions with multiplications and subtraction using Newton-Raphson Method in the CI/OFDM system. Our results confirm that with one or two iterations, the bit-error-rate (BER) performance of the CI/OFDM is achieved as the performance as of normal computation with divisions.
We study Grid computing system which provides high reliable and high accurate results of scientific simulation with ease to use. Scientists want results which are reliable in limited (shorter) time-frame. However, requirements of time and accuracy are conflict in general environments because there is a limit to the number of computer resources. In our approach, we prioritize jobs which have potential of rising good results to execute. Generally, what to determine jobs which should be prioritized is depends on kind and algorithm of simulation but it has common workflow to determine these jobs. Then we are developing the system of supporting this workflow. In this presentation, we will explain following things. 1. Design of this developing system 2. Relation between our system and method of porting scientific simulation programs to grid environments. 3. How to work our system and cooperate with other grid middleware. 4. Prototype implementation and its evaluation.
With the rapid rise in the demand for location related service, communication devices such as PDAs or cellar phones must be able to search and manage information related to the geographical location. To leverage locationrelated information is useful to get an in-depth perspective on environmental circumstances, traffic situations and/or other problems. To handle the large number of information and queries communication devices create in the current ubiquitous environment, some scalable mechanism must be required. In this paper, we propose a peerto- peer network system called “Mill” which can efficiently handle information related to the geographical location. To simplify the management of the location related information, we convert two dimensional coordinates into one dimensional circumference. Using this technique, Mill can search information by O(logN). Mill does not adopt any flooding method, and it reduces the amount of search queries compared with other P2P networks using flooding. DHTnetworks also do not leverage flooding and have good features. Simulation results show that the performance of Mill is good as well as other DHT networks. DHTssupport only exact match lookups. The exact match is not suitable for searching information of a particular region. Mill provides an effective region search, by which users can search flexibly location-related information from small regions to large regions.
Time-delay systems are encounterd in many different disciplines including engineering, biology, economics, computer science, and so on. The time-delay frequently causes instability and performance deterioration. For example, data delay in networked control system often yields unusual behavior. Therefore, stability analysis for time-delay systems has been one of the most challenging issues. In general, investigation of the stability analysis for linear time-delay systems is often performed by deriving Linear Matrix Inequality (LMI) condition which tests the stability of the system. However, such a derivation requires some matrix manipulations which are conservative operations. Then, the condition results in very conservative criterion. In this presentation, the possibility and the effectiveness of a new approach which is named Sum of Squares (SOS) is introduced. The SOS methodology is a powerful technique and generates less conservative conditions. Then, we try to apply the technique to the linear time-delay system that has control delay and feedback delay. Finally, the validity of the obtained result is demonstrated by using past example.