Collaborative Research: Retargetable Images and Video

PI将开发一种方法来转换图像，使其更有效。 这将使实用的系统能够适应为一个设备创建的图像，以便它在其他设备上有效地工作。 通过提供将图像和视频重定向到新设备的方法，PI将使应用程序能够利用激增的计算和通信基础设施。例如，将有可能向蜂窝电话提供有效的信息图像，向PDA提供训练视频，以及向“抬头”显示器提供状态信息。 为此，PI将建立一种理解和方法，以提高图像在一系列场景中的有效性，以及创建此图像的工具。这涉及开发确定源图像中重要内容的方法，改变图像以强调重要部分的方法，以及评估图像有效性的方法。 PI先前的工作表明，有效性转换可以在特定的领域中定义，并且这些可以为特定的重定向问题提供实用的工具。 在这个项目中，PI将建立在这些成功的基础上，并将其推广到更广泛的主题领域和观看场景。 该项目的成果将包括更好地理解什么是有效的图像，创建有效图像的新方法，评估图像有效性的新准则，评估图像更改的有效性，以及图像和视频重定向问题的实用系统。 正因为如此，我们的计算基础设施正在不断发展，以提供图像。 无处不在的高速网络为家庭影院屏幕、蜂窝电话显示器、联网的PDA，甚至嵌入冰箱、电梯和飞机座椅中的显示器提供图像。 不幸的是，意象只不过是一种新奇的东西，除非它是有效的，也就是说，它达到了一个交际的目的。 然而，是什么让图像有效是难以捉摸的。 这取决于主题和预期的信息。 它还取决于目标显示设备和观看情况;在墙壁大小的显示器上有效的图像可能在蜂窝电话的显示器上无效。 PI在该项目中开发的技术将有助于在各种情况下和各种设备上实现有效的图像，以便越来越普遍的显示器可以在人们需要的时间和地点为他们提供信息，更有效的人际沟通渠道，以及增强的教育媒体和娱乐体验。

The PI will develop a methodology for transforming imagery such that it is more effective. This will enable practical systems for adapting imagery created for one device such that it works effectively on others. By providing methods for retargeting images and video to new devices, the PI will enable applications to make use of the proliferating computing and communications infrastructure. For example, it will become possible to provide effective informational images to cellular phones, training videos to PDAs, and status information to "heads up" displays. To these ends, the PI will establish an understanding and methodology for transformations that improve the effectiveness of imagery across a range of scenarios, and tools to create this imagery. This involves developing methods for determining what is important in the source imagery, methods for altering images to emphasize the important parts, and methods for evaluating the effectiveness of imagery. The PI's prior work shows that effectiveness transformations can be defined in specific domains and that these lead to practical tools for specific retargeting problems. In this project, the PI will build on and generalize these successes to a broader range of subject domains and viewing scenarios. The outcomes of this project will include a better understanding of what makes effective imagery, new methods for creating effective imagery, new guidelines for assessing the effectiveness of imagery, evaluations of the effectiveness of image alterations, as well as practical systems for image and video retargeting problems.Broader Impacts: Images can be a powerful mode for communication. Because of this, our computing infrastructure is evolving to deliver imagery. Ubiquitous high-speed networking provides imagery to home theater screens, cellular phone displays, networked PDAs, and even displays embedded in refrigerators, elevators and airplane seats. Unfortunately, imagery is little more than a novelty unless it is effective, that is, it achieves a communicative goal. Yet what makes imagery effective is elusive. It depends on subject matter and intended message. It also depends on the target display device and viewing situation; images effective on a wall-sized display may be ineffective on the display of a cellular phone. The techniques to be developed by the PI in this project will help make effective imagery a reality in a wide variety of situations and for a wide variety of devices, so that increasingly ubiquitous displays can provide people with information when and where they need it, more effective channels of inter-personal communication, and enhanced educational media and entertainment experiences.