No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Fisheye videos: distorting multiple videos in space and time domain according to users' interests
Abstract
Many applications, such as tele-conference systems and plant control systems need to display a large number of videos. In those applications, displaying multiple video windows overwhelms limited computing resources (e.g., network capacity, processing power) due to the vast amount of information. This paper describes a technique allows multiple videos to display in the limited computing resources. This technique distorts multiple videos according to users’ interest. Users are not interested in all videos simultaneously. They only look at a part of them in detail and get the global context of other videos. The technique displays videos of interest in more detail by degrading other videos to allow an efficient use of limited computing resources, which we call the Fisheye Videos technique, The technique distorts a video in the space and time domain (e.g., spatial resolution, frame rate) according to users’ interests, which are estimat@ based on the window conditions such as its distance from a focused window and the amount of masked area by other windows.
... Une autre technique appelée Fisheye Videos avaitété proposé dès 1993 par Yamaashi et ses collègues pour la présentation de plusieurs vidéos simultanées [Yamaashi et al. 1993]. Elle répondait surtout au problème de l'époque concernant les limites des ressources de calcul et de mise en mémoire de nombreuses vidéos. ...
Although more and more sound and audiovisual data are available, the majority of access interfaces are solely based on a visual presentation. Many visualization techniques have been proposed that use simultaneous presentation of multiple documents and distortions to highlight the most relevant information. We propose to define equivalent audio technique for the presentation of several competing sound files, and optimally combine such audio and visual presentation strategies for multimedia documents. To better adapt these strategies to the user, we studied attentional and perceptual processes involved in listening and watching simultaneous audio-visual objects, focusing on the interactions between the two modalities.Combining visual size and sound level parameters, we extended the visual concept of magnifying lens to auditory and audiovisual modalities. Exploiting this concept, a navigation application in a video collection has been developed. We compared our tool with another rendering mode called Pan & Zoom through a usability study. Results, especially subjective results, encourage further research to develop multimodal presentation strategies by combining an audio rendering to the visual renderings already available.A second study concerned the identification of environmental sounds in a noisy environment in the presence of a visual context. The noise simulated the presence of multiple competing sounds as would be observed in an interface where several multimedia documents are presented together. The experimental results confirmed the multimodality advantage in condition of audio degradation. Moreover, beyond the primary goals of the thesis, this study confirms the importance of the semantic congruency between visual and auditory components for object recognition and provides deeper knowledge about the auditory perception of environmental sounds.Finally, we investigated attentional processes involved in the search of a specific object among many, especially the “pop-out” phenomenon whereby a salient object automatically attracts attention. In vision, an sharp object attracts attention among blurred objects and some visual strategies already exploit this parameter to display the information. We extended by analogy the concept of visual blur to auditory and audiovisual modalities. A serie of experiments confirmed that a perceptual object among blurred objects attracts attention, regardless of the modality. The identification and search process is then accelerated when the sharpness parameter is applied to the target, but slow when it is applied to a distractor. These results highlight an involuntary attraction effect. Concerning the crossmodal interaction, a redundant combination of audio and visual blur proved to be more effective than a unimodal presentation. Results also indicate that optimal combination does not necessarily require a distortion of both modalities.
We describe two systems that use interactive computer-controlled video for shared awareness and casual communication. Polyscope lets users monitor a large number of video sources simultaneously. Observers are provided with a window containing a collection of framegrabbed bitmap images or animations. These images can be used to access additional video services, such as videophone. Vrooms is a follow-on system, which employs a strong spatial metaphor. Users can enter and leave virtual rooms. Once in a virtual room, users can see and be seen by all the other occupants, and have easy access to other video, audio, and text-based communication tools.