项目名称: 云环境中基于三维世界模型的图像表示与压缩
项目编号: No.61303149
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 自动化技术、计算机技术
项目作者: 刘东
作者单位: 中国科学技术大学
项目金额: 23万元
中文摘要: 云环境中海量的图像/视频数据和云所要支持的丰富应用之间的矛盾给图像表示与压缩提出了新的挑战,而三维世界模型的逐步建立则为有效地表示自然图像提供了新的机遇。本项目拟研究的关键科学问题是在云计算环境中如何有效地表示和压缩图像/视频数据,以便与其他类型的数据有机地关联、整合,采取的研究途径是基于三维世界模型和海量图像库,提取、描述图像中的关键视觉信息,并利用基于模型的和基于图像的绘制技术,根据关键视觉信息重建原图像。本项目的研究内容包括:1) 一种普适的基于模型的图像编解码框架,利用三维模型和图像库作为先验知识;2) 基于三维模型和图像库的图像分析和合成方法,进行图像和关键视觉信息的相互变换;3) 对所提图像编解码框架的评判和优化,综合考虑压缩性能、计算性能、数据组织和管理能力等。本项目的实施将为云计算中的图像/视频数据表示提供实用的方案,并推动基于模型的图像编码进一步发展。
中文关键词: 图像表示;图像压缩;三维模型;基于模型的编码;图像标注
英文摘要: How to organize and manage the huge amount of image/video data in the cloud so as to support various applications is a new challenge to the image representation and compression research. Meanwhile, the gradually built 3D world model provides opportunities for efficiently representing natural images. Motivated by these observations, in this project, it is planned to investigate how to efficiently represent and compress image/video data in the cloud and make the data connected and integrated with other kinds of data. The planned approach consists of extracting and describing the distinctive visual information in an image with the help of 3D world model as well as a huge library of related images, and reconstructing the image from the coded distintive visual information by means of model-based and image-based rendering. Research issues in this project include, 1) a universal model-based image coding/decoding framework that leverages related 3D models and images as prior knowledge; 2) image analysis/synthesis methods based on both 3D model and image library, performing the forward/inverse transform between an image and its distinctive visual information to be coded; 3) the assessment and optimization of the proposed image coding/decoding framework, taking into account the compression performance, computation perform
英文关键词: image representation;image compression;3D model;model-based coding;image annotation