项目名称: 基于无穷范数和2范数联合优化的低复杂度高保真图像编码研究
项目编号: No.61301288
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 无线电电子学、电信技术
项目作者: 牛毅
作者单位: 西安电子科技大学
项目金额: 28万元
中文摘要: 在卫星通信、远程医疗等应用中,图像编码不仅仅需要考虑整体失真,还需要控制单个像素的最大绝对误差。此外,受采集设备和带宽的限制,编码复杂度和码率控制同样重要。然而传统编码算法均在上述某方面存在缺陷,无法同时满足要求。 考虑到最大绝对误差和整体失真分别可由L∞和L2范数度量,本项目将结合申请人前期工作,提出基于L∞,L2 联合优化的新型编码框架,并对若干关键问题进行深入研究。在编码复杂的控制方面,本研究将从高性能解码恢复入手,重点研究图像支撑域、阶数自适应建模及相关优化算法,有望提出L∞约束L2最优解码算法,完成复杂度从编码端向解码端转移;在码率控制方面,本项目提出残差渐进编码机制,重点研究L∞约束渐进量化算法,有望实现L∞约束低复杂度最优步长分配,精确码率控制。 本项目提出的编解码框架不仅有着重要的实际意义,研究成果还将对自回归建模、分布式编码、最优量化器设计等基础理论的发展产生推动作用。
中文关键词: 图像压缩;图像重构;自回归模型;图像增强;高动态压缩
英文摘要: The transfer of satellite image and medical image is more critical than natural images. Both the overall distortion and peak absolute error should be considered . In addition, the encoder is limited by power consumption and transfer bandwidth. The transfer system should also maintain a low encode complexity and offer flexible rate control mechanism. However, the existing coding techniques are hard to meet the above requirements. Considering that the overall distortion and peak absolute error can be characterized by L∞ and L2 norm respectively, in this proposal we investigate a novel joint L∞,L2 optimized coding system. Instead of ?ne tuning the encoder, the new system move the task of improving the coding efficiency to the decoder. Depending on the research of adaptive modeling of decoded images, we will propose an optimum L∞ constrained L2 decoding technique to restore the code stream. Another novelty of this proposal is the flexible L∞ constrained rate control mechanism. We will propose a new low complexity L∞ constrained residual coding mechanism which provides fine grain rate control via L∞ constrained scalable quantization. The proposed system is not only an improved practial will achieve higher coding performance than the current coding techniques while still maintain the low. , near lossless coding with
英文关键词: Image compression;Image reconstruction;Piece-wise autoregressive model;Iamge enhancement;High dynamic range compression