Contents Enforme: Automatic Deformation of Content for Multi-features without Information Loss

摘要

We introduce a deformation technique that enlarges the feature areas in an image while retaining the information in the non-feature areas. Our main purpose is to provide an effective thumbnail that is useful for practical use in small display devices (e.g., cellular phones, digital cameras, and game devices). Even though many approaches to achieve our purpose have been developed, they are not useful because they require enough time to calculate features. In contrast, our approach can quickly deform image features based on a rapid segmentation technique that our laboratory has proposed. Moreover, our system calculates each segmented area, so it can treat multi-features as deformation elements. As a result, the feature areas were enlarged and the non-feature areas were reduced at almost processing speed, and the total information contained in the original image was retained. Our smooth deformation technique is useful not only for image deformation, but also for a wide variety of contents such as net-meeting and video contents. In this paper, we describe the concept underlying the image enforme technique and its applications.