Image Processing Laboratory

Ryuichi Oka Professor	Keitaro Naruse Assistant Professor	Yu Nakajima Research Associate

First and foremost, the Image Processing Laboratory engages in research and development of image-based pattern recognition including some areas of Artificial Intelligence, database organisation, retrieval, and robotics. More specifically, as can be seen from the background information and the recent research publications of the members of the laboratory, object tracking in a time-varying image, image retrieval from video data, character recognition of telop in a video image and extracion of the Internet community are our current focus. Related to the recent progress of the web, a huge amount of multimedia data without index becomes available to store in our PC's. However, no sophistcated methodology to manage such data has been developed so far. So that we are solicited to attach an index to each data. Our research aim is to develop algorithms to realize automatic annotation to real word data for integrated retrieval of multimedia information. The algorithms include self-organisation and transformation among representation of multimedia and feature extraction and recognition of real data. Real world data includes video, still image, speech, music, sound, and text each of which has not been indexed by labels. An example of multimedia integration retrieval called "CrossMediator" developed by ten-year project (1992-2002) RWC of METI (Japan). Some parts of CrossMediator have been in the commercial market by through a private company. Prof. R.Oka was a chief of group which developed "CrossMediator". Our laboratory will pursue to develop more sophisticated functions which might reveal a new generation of the Internet. Speech retreival system used in CrossMediator is used for developing new types of commercial products. One of the recent progress is developing a set of algorithms for spotting recognition and tracking objects in a time-varying image. These algorithms enable to realize automatic anonotation of video image capturing moving objects. Another recent progress is developing a spotting algorithm for song wave retrieval. A large amount of song wave data is available for anyone along with the progress of storage hardware of music data. But the progress of technology for accessing the song wave data seems quite slow so that we must use conventional and not sophistcated tools. Our new software can provide users a convenient way to find and rerieve the song wave data by singing an arbitrary part of it which he/she wants to hear. The Internet is regarded as a network composed of virtual communities. Visualization of the community is becoming an important research target. Our lab is developing a sophistcated algorithm based on so-called Associated hyper-linked word space (A-space). We use the algorithm to detect clusters each of which corresponds a virtual community. Visualization of each virtual community is realized by showing the content extracted from the web pages included the community.

Refereed Journal Papers

[oka-01:2005]

Hiroshi Kameya, Shunji Mori, and Ryuichi OKA. A Segmentationfree Biometric Writer Verification Method Based on Continuous Dynamic Programming. Pattern Recognition Letters, 27(6):567-577, 2006.

The paper proposed a sophisticated method for writer verification. The key contribution includes the followings: segmentation-free verification, languagefree, applicable to both written character sequence and written figure. Segmentation-free function allows users to write an arbitrary part of registered reference. Language-free function provides a wide application area. Applicability of figure pattern provides also a wide application area of security. The proposed method was compared with the best conventional method and obtained a higer score.

Refereed Proceeding Papers

[naruse-01:2005]	K. Naruse and M. Oya. Muscle-like Control of Entertainment Robot over Internet. In Proc. of the 2005 IEEE/ASME International Conference on Advanced Mechatronics (AIM2005), pages 1419-1424, Jul. 2005.
	The robot technology is extending to social activities such as community work, home-use and entertainment. By combining the robot technology with the popularized public Internet, it is anticipated a new application field will come out. However, Internet, as its nature, does not permit real-time communication among robots and other systems including human operators and computers. A different approach is necessary regarding communication protocols and robot control schemes. This paper proposes the combination of the Web services technology and the autonomous robot motion generation is the proper approach. For the motion generation, we introduce the muscle-like control method, which can be applied to any kinds of robots easily with proving a variety of creature-like motions. The method is applied to gait generation of a simulated four-legged robot, and the results show that it can transit from a motion to another one smoothly. In addition, the simulated robot is controlled by the Web services over Internet for investigating the feasibility of the proposed control scheme in a real network circumstance. The result shows that the proposed method provides us smooth control of the robot, even we have overhead of Web service processes and delayed communication.
[naruse-02:2005]	K. Naruse, S. Kawai, and T. Kikuchi. Three-dimensional Liftingup Motion Analysis for Wearable Power Assist Device of Lower Back Support. In Proc. of the 2005 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS2005), pages 3126-3131, Aug. 2005.
	The objective of this research is to develop a wearable power assist device which helps a person to lift up a heavy object. The device is designed to support him by holding his upper body weight and reducing his inner force. It turns to the reduction of a compression force of his lower back discs, which is a major factor of a lower back injury. In a lifting-up motion, he twists and bends his body, which means three-dimensional motion analysis is necessary for the compression force analysis. However, most of related works on the compression force analysis have been carried out in a vertical plane. For understanding the characteristics of a lifting-up motion better, this paper presents three-dimensional analysis of it. Using a motion capture system, we measured a sequence of body positions with and without the power assist device in different object location and mass, as well as surface EMG (electormyogram) signals. The compression force of the lower back discs is estimated by a three-dimensional biomechanical human body model utilizing the position data. The results show that the range of the twist angle of his body is about 30 degree in maximum and the compression force increases about 10twist angle. With respect to the effect of the power assist device, the magnitude of the surface EMG signals of spine muscles is decreased when wearing the power assist device. It means the proposed power assist device gives a good support to him.
[naruse-03:2005]	K. Naruse and R. Oka. Search in Linked Document Space by Social Topology Agents. In Proc. of the 2005 IEEE Congress on Evolutionary Computation (CEC2005), pages 850-855, Sept. 2005.
	When we use a search engine in Internet, we often cannot retrieve a document we look for, due to the difficulty of finding adequate keywords. One of the ways to solve the problem is to search adequate keywords interacting with a user: A metasearch engine displays keywords and pages, a user evaluate them. Iterating the process, the metasearch engine narrows the keywords, getting closer to what a user looks for. For realizing it, we construct a graph called a linked document space, in which nodes and links represent the pages and the similarities, respectively. Then, the metasearch engine searches every promising page considering relations between pages, following the evaluation from a user. This paper presents a search method in the linked document space, called the socially topology agents (STA), which are inspired by human social relation and swarm intelligence. STA is applied to a multi-peak largescale search problem, and it is shown that STA can find most of peaks quickly, as well as the aggregation and distribution of the agents in the space are controlled by a user command. They mean STA is feasible as the metasearch engine.
[naruse-04:2005]	M. Kubo, K. Naruse, and H. Sato. Behavior of Chat Group inspired by BTW Model. In Proc. of 2005 Japan-Australia Workshop on Intelligent and Evolutionary Systems (IES2005), Nov. 2005.
	In this paper, we discuss characteristics of etext-basef communicaton via internet using simulator of Bak, Tang, and Wiesenfeld model (namely BTW model as you know.) And then, the simulator can demonstrate very similar behavior to be in real.
[naruse-05:2005]	Y. Yaguchi, H. Ohnishi, S. Mori, K. Naruse, R. Oka, and H. Takahashi. A Mining Method for Linked Web Pages Using Associated Keyword Space. In Proc. of the 2006 IEEE-CS/IPSJ International Symposium on Applications and the Internet (SAINT2006), pages 268-276, Jan. 2006.
	We propose a novel method for mining knowledge from linked Web pages. Unlike most conventional methods for extracting knowledge from linked data, which are based on graph theory, the proposed method is based on our Associated Keyword Space (ASKS), which is a nonlinear version of linear multidimensional scaling (MDS), such as Quantification Method Type IV (Q-IV). We constructed a threedimensional ASKS space using linked HTML data from the World Wide Web. Experimental results confirm that the performance of ASKS is superior to that of Q-IV for discriminating clusters in the space obtained. We also demonstrate a mining procedure realized by 1) finding subspaces obtained in terms of logical calculations between subspaces in an ASKS space and 2) detecting emerging spatial patterns with geometrical features.
[oka-02:2005]	Yuuichi Yaguchi, Hiroshi Oonishi, Satoshi Mori, Keitarou Naruse, Hironobu Takahashi, and Ryuichi Oka. A Mining Method for Linked Web Pages Using Associated Keword Space. In International Symposium on Applications and the Internet(SAINT2006), pages 268-276. IEEE, Jan. 2006.
	We propose a novel method for mining knowledge from linked Web pages. Unlike most conventional methods for extracting knowledge from linked data, which are based on graph theory, the proposed method is based on our Associated KLeyword Space (ASKS), which is a non-linear version of linear multidimensional scaling (MDS), such as Quantification Method Type IV (Q-IV). We constructed a three-dimensional ASKS space using linked HTML data from the World Wide Web. Experimental results confirm that the performance of ASKS is superior to that of Q-IV for discriminating clusters in the space obtained. We also demonstrate a mining procedure realized by 1) finding subspaces obtained in terms of logical calculations between subspaces in an ASKS space and 2) detecting emerging spatial patterns with geometrical features.
[oka-03:2005]	Yuuichi Yaguchi and Ryuichi Oka. Song Wave Retrieval Based on WFrame-Wise Phoneme Recognition. In Second Asia Information Retrieval Symposium(AIRS 2005), pages 503-509. ACM, Oct. 2005.
	We propose a song wave retrieval method. Both song wave data and a query wave for song data are transformed into phoneme sequences by frame-wise labeling of each frame feature. By applying a search algorithm, called Continuous Dynamic Programming (CDP), to these phoneme sequences, we can detect a set of similar parts in a song database, each of which is similar to a query song wave. Song retrieval rates hit 78 percent in four clauses from whole databases. Differences in each query from song wave data and speech wave is investigated.

Unrefereed Papers

[naruse-06:2005]	K. Naruse and R. Oka. Recognition of Bone Conducted Voice Commands by Continues Dynamic Programming. In JSME Conference on Robotics Mechatronocs 2005, Jun. 2005.
	One of the major difficulties of speech recognition in a daily life circumstance is miss recognition due to an environment noise. In this paper, we introduce a bone conduction microphone, which is robust to the noise because of its voice picking up mechanism. However, the bone conduction microphone has the disadvantage of a narrow bandwidth, and it makes speech recognition of the bone conducted voice difficult. Fore improving the accuracy of the speech recognition of the bone conducted voice, in this paper, we introduce a continuous dynamic programming (CDP) method, which is capable of extending or shrinking of an input pattern, as an algorithm for matching an input voice with a template one. The experimental results show that the proposed method works well for the bone conducted voice.
[naruse-07:2005]	Kikuchi T., K. Naruse, S. Kawai, and S Yoshii. Development of Wearable Power Assist Device - Load Analysis of Lower-back Flexion, Extension and Rotation -. In JSME Conference on Robotics Mechatronocs 2005, Jun. 2005.
	The objective of this research is to develop a wearable power assist device which helps a person to lift up a heavy object. The device is designed to support him by holding his upper body weight and reducing his inner force. It turns to the reduction of a compression force of his lower back discs, which is a major factor of a lower back injury. In a lifting-up motion, he twists and bends his body, which means three-dimensional motion analysis is necessary for the compression force analysis. However, most of related works on the compression force analysis have been carried out in a vertical plane. For understanding the characteristics of a lifting-up motion better, this paper presents threedimensional analysis of it using a motion capture system. We propose a human model, and estimate the compression force of the lower back discs based on the body position data with and without the power assist device. The results show that a decrease in the lower back load by power assist device.
[naruse-08:2005]	K. Naurse. Three-dimensional Lifting-up Motion Analysis for Wearable Power-assist Device for Lower-back Support. In 2005 Fall JSPE Conference, Sept. 2005.
	The objective of this paper is to understand human motion, particularly lifting-up motion, for controlling a wearable power assist device. Conventional studies on the motion analysis only consider two-dimensional motions assuming symmetrical motion, although actual human motions are threedimensional. This paper presents the analysis of three-dimensional position data of human lifting-up motions are acquired by an optical motion capturing system.
[naruse-09:2005]	K. Naruse. Remote Operation of Entertainment Robot by Web Service and Virtual Muscle Oscillation Method. In 2005 SICE System Integration Division Conference (SI2005), Dec. 2005.
	The objective of this paper is to develop a method for operating an entertainment robot remotely, easy even for a beginner user, by providing a same robot operation and communication interface for any kinds of robots. In this paper, we introduce virtual muscle oscillation method for robot motion control, which is applicable to any existing robots with small steps, and the web service technologies for communication, which is widely used and easily extensible to Web servers. . Experimental results show that the proposed method works well for a simulated four-leg robot.
[naruse-10:2005]	K. Naruse. and H. Onodera. Effect of Small World Property to Search Ability in Socially Topology Agents. In 2006 Spring JSPE Conference, Mar, 2006.
	The objective of this paper is to investigate a search ability of socially topology agents, which is inspired by social relation of humans. On the other hand, it has been known that a group of human often forms a social relation of a graph structure called small world. The small world graph can be generated from a regular graph by adding some random rewiring, so it can represent a wide range of random graphs. In this paper, we investigate how much of randomness provides good search ability by computer simulations. The results show that small randomness shows a good search performance in a spiky multimodal function with less number of agents.

Grants

[naruse-11:2005]

K. Naruse. Grant-in-aid, Exploratory Research, 2004-2006.

Academic Activities

[naruse-12:2005]	K. Naruse, 2005. Organizer, 2005 SICE System Integration Division Conference
[naruse-13:2005]	K. Naruse, 2005. Reviwer, International Journal of General Systems
[naruse-14:2005]	K. Naruse, 2005. Reviwer, JSAI
[naruse-15:2005]	K. Naruse, 2005. Reviwer, IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS2005)
[oka-04:2005]	Ryuichi Oka, May 2005. Reviewer of submitted papers, The Acoustic Society of Japan
[oka-05:2005]	Ryuichi Oka, May 2005. Reviewer of submitted papers, The Institute of Electronics, Information and Communication Engineers
[oka-06:2005]	Ryuichi Oka, April 2005. Member of editorial committee, Japanese Society for Artificial Intelligence
[oka-07:2005]	Ryuichi Oka, April 2005. Reviwer of submitted papers, the Six Inter. Conf. on Computer and Information Technology

Ph.D and Other Theses

[naruse-16:2005]	Syun Andou. Graduation Thesis: Three-dimensional Operation Generation from Two-dimensional Animation, University of Aizu, 2005. Thesis Advisor: Naruse, K.
[naruse-17:2005]	Hayato Onodera. Graduation Thesis: A Search Method Using Particle Swarm Optimization and Small World Structure, University of Aizu, 2005. Thesis Advisor: Naruse, K.
[naruse-18:2005]	Tatsuya Kato. Graduation Thesis: Preferred Behavior Generation by Interation with Users, University of Aizu, 2005. Thesis Advisor: Naruse, K.
[naruse-19:2005]	Masayo Kobayashi. Graduation Thesis: Artificial Evolution for Morphology of Virtual Creatures, University of Aizu, 2005. Thesis Advisor: Naruse, K.
[naruse-20:2005]	Makiko Tashima. Graduation Thesis: Statistical Analysis of Facial Expression Combinations, University of Aizu, 2005. Thesis Advisor: Naruse, K.
[naruse-21:2005]	Eisuke Mayuzumi. Graduation Thesis: Remote-controlled Robot using Web-service and Virtual Muscle Oscillation Method through Internet, University of Aizu, 2005. Thesis Advisor: Naruse, K.

Others

[oka-08:2005]	Ryuichi Oka, Hironobu Takahashi, Takuichi NIshimura, and Toshiro Mukai.
	An artcle titled Realtime Gesture-Speech Interface Realized on a Personal Computer of Notebook Size (pages: 206-215) is included in a book titled Spoken Language Systems edited by Seiichi Nakagawa and Michio Okada and Tatsuya Kawabata (Ohmsha,2005).

2007 THE UNIVERSITY OF AIZU ALL RIGHTS RESERVED