项目名称: 基于压缩感知理论的视频编解码技术研究
项目编号: No.61471343
项目类型: 面上项目
立项/批准年度: 2015
项目学科: 无线电电子学、电信技术
项目作者: 沈燕飞
作者单位: 中国科学院计算技术研究所
项目金额: 82万元
中文摘要: 随着压缩感知理论的逐步成熟,为突破以奈奎斯特采样定理为基础的信号处理框架提供了新的理论支撑,给视频图像信号的采集、编码、存储和处理等带来了前所未有的突破。本项目将研究基于压缩感知理论的视频编解码技术,将视频采集和视频编码进行深度融合,构建一套完整的压缩感知视频编解码技术框架体系,为压缩感知理论在视频编解码方面的应用提供理论支持和算法保证。主要研究内容包括:在视频感知测量方面,采用近似确定性感知测量矩阵,基于自适应感知测量模型对视频信号进行感知测量,利用感知域中的相关性对测量值进行预测、量化和熵编码;在视频图像重建方面,利用感知域中的时空相关性,研究一种基于预测残差的压缩感知视频重建算法;为了提高视频编码的率失真性能,还将对感知测量矩阵和稀疏字典进行联合优化,并使用序列化压缩感知技术来实现自适应压缩感知视频编码。通过本项目的研究将为压缩感知理论在视频编解码方面的应用提供技术参考和理论指导。
中文关键词: 视频信息采集;视频重建;多媒体信息处理
英文摘要: With the rapid development and gradually mature of compressive sensing (CS) theory, it provides strong theory support for new signal processing framework beyond the Nyquist sampling theorem, and thus represents an unprecedented breakthrough in many fields such as video sampling, video coding, video storage and video processing. In this project, the video coding technology based on CS theory will be studied, which deeply fuses video sampling and coding process, and a new technique framework of video coding based CS theory will be provided in order to extend the application of CS theory in the field of video coding. The main research contents include: in the terms of video CS measurement, Video sequences are sensed based on adaptive CS measurement model by use of approximate determined measurement matrix, and the sensed data are predicted, quantization and entropy coding in measurement field; in the term of video reconstruction, the prediction residue will be used to reconstruct the video based the spatial and temporal relation between successive video frames. In order to improve the efficiency of video coding, CS measurement matrix and sparse representation matrix will be optimized jointly, and sequential CS technology will be used to improve the network adaptive for our proposed CS video codec. In summary, this project will provide technology reference and theoretical guidance for the design of video coding framework based on compressed sensing theory.
英文关键词: Video Sampling;Video Recovery;Multimedia processing