大澤 範高

視覚の補完機能であるアモーダル補完を応用し，人間には負担が大きすぎず，自動文字認識には攻撃コストが増加し難度が高いCAPTCHA（Completely Automated Public Turing test to tell Computers and Humans Apart）を生成する手法を提案し，評価を行った．提案手法では背景色の図形によって文字の一部を欠けさせた欠損画像と，欠損画像上の欠けた部分を覆い隠すことができる図形を透明背景に描画したマスク画像の2枚を提示する．欠損部にマスクがかかるように2枚の画像を重ね合わせると，アモーダル補完の効果により人間にとっては容易に文字認識ができる．画像の重ね合わせ操作は，ボットが攻撃に必要とするコストを増大させると期待できるが，欠損画像のみから文字認識されると効果がない．そこで，3種類の認識困難化手法を組み合わせた欠損画像を生成し，畳み込みニューラルネットワークによる自動文字認識率を評価した．反転ノイズ重畳と文字幅・間隔不均一化の組合せが最も効果があり，正解率を認識困難化手法を使わない場合の0.946から0.788に低減可能であることを示した．また，被験者による文字列読み取り実験によって，提案したCAPTCHAに対する正解率と解答時間および操作に対する主観的な負荷について調査した．画像の重ね合わせを適切に行うことで，アモーダル補完の効果によって文字の認識が容易になることを確認し，画像の重ね合わせ操作に対する負担感軽減が必要であることも明らかにした．

A two-handed distance control method is proposed for precisely and efficiently manipulating a virtual 3D object by hand in an immersive virtual reality environment. The proposed method enhances direct manipulation by hand and is used to precisely control and efficiently adjust the position of an object and the viewpoint using the distance between the two hands. The two-handed method is evaluated and compared with the previously proposed one-handed speed control method, which adjusts the position of an object in accordance with the speed of one hand. The results from experimental evaluation show that two-handed methods, which make position and viewpoint adjustments, are the best among six combinations of control and adjustment methods.

Three-dimensional visualization using jigsaw-puzzle-like glyphs, or shapes, is proposed as a means of representing grammatical constraints in programming. The proposed visualization uses 3D glyphs such as convex, concave, and wireframe shapes. A semantic constraint, such as a type constraint in an assignment, is represented by an inclusive match between 3D glyphs. An application of the proposed visualization method to a subset of the Java programming language is demonstrated. An experimental evaluation showed that the 3D glyphs are easier to learn and enable users to more quickly understand their relationships than 2D glyphs and 1D symbol sequences.

Collaboration is important to improve education. NIMEfNational Institute of Multimedia Education) in Japan has built up an inter-university collaboration network, SCS(Space Collaboration System), using satellite technologies. It provided natural collaboration environment using two-way 1.5 Mbps channels and has been operated by NIME since 1996. It gave easy multi-site and multi-channel communication capability to 150 sites in 123 universities, colleges and other centers, Many collaboration activities have been performed on SCS for a total of more than 23,000 hours, 9,800 sessions and 157,000 site*hours. More than half of the sessions were with more than three sites. Half of the sessions were for joint classes. And this satellite system was used more frequently in thinly populated area than the densely populated area. A classroom of students could join important outside activities and the interaction with outside people gave them motivation to study. SCS is thought to have contributed to the improvement of education. This article reports the operational results of the inter-university collaboration system for more than 10 years. Copyright © 2006 by the International Astronautical Federation. All right reserved.

Two-handed control techniques for precisely and efficiently manipulating a virtual 3D object by hand in an immersive virtual reality environment are proposed. In addition, one-handed and two-handed techniques are described and comparatively evaluated. The techniques are used to precisely control and efficiently adjust an object with the speed of one hand or the distance between both hands. The controlled adjustments are actually position and viewpoint adjustments. The results from experimental evaluations show that two-handed control methods that are used to snake the position and viewpoint adjustments are the best, but the simultaneous use of both one-handed and two-handed control techniques does not necessarily improve the usability.

Automatic adjustment methods for efficient, precise positioning and release of a virtual 3D object by direct hand manipulation in an immersive virtual reality environment are described and evaluated. The proposed methods are release adjsutment, position adjustment, viewpoint adjustment, and virtual hand size adjustment. Combining these methods enables users to Manipulate a virtual object efficiently and precisely. An experimental evaluation showed that these methods were effective and useful in terms of the number of task completions and the subjective preference, particularly for a small virtual target.

We developed a prototype system for graphical drawings on Web browsers using Asynchronous JavaScript and XML (AJAX) techniques. The system allows multiple participants to share multiple pointers and to simultaneously add graphical drawings onto a Web page. All the participants can see all telepointers and graphical drawings in real-time. They only need a normal Web browser and installation of special software such as plugins is not required to share the telepointers and graphical drawings. Since the Web browser communicates with a server through conventional HyperText Transfer Protocol (HTTP) connections, telepointers and graphical drawings can be widely shared without communications being prevented by network firewalls. Various methods of optimization are also discussed in the paper.

We developed an interactive system of displaying personalized information that uses a combination of a half mirror and an LCD display. The LCD display can provide information and the half mirror can be used as a normal mirror when information is not being displayed. Replacing existing mirrors with these systems also saves space. Personalized information is displayed when a user's face is recognized using a camera to identify the individual. Camera-based face recognition does not require contact and is not intrusive. Moreover, a touch panel on the half-mirror display enables users to select items to find detailed information. Thus, the system is not only a half-mirror display but it is also interactive. An evaluation based on questionnaires was conducted, which clarified advantageous locations and authentication methods, as well as customized information that is appropriate and necessary to display.

Immersive virtual reality (VR) has long been considered an excellent environment in which to manipulate 3D virtual objects. However currently used immersive VR user interfaces have limitations. For example, while direct manipulation by hand is easy to understand and to use for approximate positioning, direct manipulation by hand is not suitable for making fine adjustments to virtual objects in an immersive environment because it is difficult to hold an unsupported hand in midair and then to release an object at a fixed point. We therefore propose a method that combines direct 3D manipulation by hand with a virtual 3D gearbox widget that we recently designed. Using this method, hand manipulation is used first to move virtual objects and place them in an approximate position, and then the widget is used to move them into a precise position. The experimental evaluation showed that this combination of direct manipulation by hand and the proposed gearbox is the best of five tested methods in terms of completion ratio of task and subjective preference.

We developed a multimedia conferencing system using IPv6/IPv4 multicasting for real-time distance education that enables video, audio, and pointing information to be shared. It supports multiple telepointers that provide group awareness information and makes it easy to share interests and attention. In addition to pointing with the telepointers, it allows users to add graphical annotations onto video and share these with one another. It also supports attention sharing using video processing techniques. The system includes a sender application, a receiver application, a multipoint viewer, a recording/replaying/relaying tool and a voting application, along with the remote control tools for these. A modular design enables the system to adapt flexibly to various configurations with different numbers of displays and different resolutions at multiple sites. Moreover, the remote control applications help the chairperson or conference organizer to simultaneously change the settings of a group of applications distributed at multiple sites.

In this paper, we first ascertain by experiment, which collects the statistics of UDP packets exchanged between two IP-based video conference systems, that the packet loss rate of an IP network over a satellite link used in Space Collaboration System (SCS) is almost equal to zero and much smaller that that of the Internet. Next, we observe deterioration of received video image caused by lost packets. This observation suggests that lost packet mainly causes an interruption of video image. The results of above experiments suggests that the IP network over SCS is more suitable to IP-based video conference than the Internet.

We have developed a prototype system for 3D video distance education using satellite communications and IP multicasting. Our system is based on personal computers and software codecs. The system is inexpensive and flexible for various configurations because it utilizes software modules. Our system can be easily installed outdoors since it does not use special hardware to encode, decode, or transmit multimedia data. Testing of our system implemented between a horticulture farm and an indoor studio, approximately 140 km apart, showed that it effectively supports distance education between indoor and outdoor environments and that it provides an enhanced 3D educational environment. © 2005 IEEE.

We have developed a personal 3D videoconferencing system. A sender using the system can capture stereoscopic images from a stereo camera, encode them, and transmit them. The receiver can then receive the compressed data, decode it, and render the decoded images on a stereoscopic display. 3D video is useful for understand natural environments or scenes in which depth is important. The system mainly utilizes an autostereoscopic display, which is easy to use because it does not require glasses for stereoscopic views. The system enables users to share 3D telepointers, and it supports the creation of 3D graphical annotations by using telepointers, or by line drawing in 3D space. Finally, the system utilizes IPv4/IPv6 multicasting to exchange stereoscopic video, audio, and shared telepointer information among multiple sites.

We have been developing an interactive and realtime multimedia distance education system among multiple sites using IP (Internet protocol) -based satellite networks. Video, audio, telepointers, and other data must be efficiently shared among many sites through such a network. We have therefore designed basic multicast-capable networks for efficient communication through satellite communication. We have developed a set of tools, called FocusShare, which include a sender application, a receiver application, a multipoint viewer, and other tools. Using two satellite communication networks and the developed tools, we have conducted experiments in cooperation with groups in other countries to demonstrate the appropriateness of our satellite IP network, and have shown that our tool suite is useful for point-to-point and multipoint conferencing.