项目名称: 基于随机相位调制的自然场景压缩成像方法与实现研究
项目编号: No.61501001
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 无线电电子学、电信技术
项目作者: 张成
作者单位: 安徽大学
项目金额: 23万元
中文摘要: 压缩成像是压缩感知理论最核心应用领域之一,比经典成像模式在成本、便携性、功率、高分辨率等方面都具有重要的潜在优势,但现有压缩成像方法或存在曝光时间太长,或实现成本过高,或需要大量标定工作等问题。前期我们已围绕压缩成像中测量矩阵的设计做了一些先行研究,验证随机相位调制压缩成像方法的可行性,以及低实现成本的确定性相位掩膜调制的有效性。本项目拟在前期研究基础上进一步探讨自然光照明条件下的复光场经过频域相位随机调制的压缩成像模型,建立复光场与可直接被光电器件记录的强度信息之间的对应关系;同时基于最适合描述衍射复光场的Fresnelets小波与光场的特殊结构,构建更紧致的结构化稀疏模型,探讨自然场景物理约束和结构化稀疏模型对重建算法的影响,针对复光场的振幅和相位之间互相耦合性,通过改进经典重建算法从真实的测量数据精确地重建原对象。该研究成果可以为新一代成像设计在理论、计算和技术上提供新的借鉴和支撑。
中文关键词: 压缩感知;压缩成像;结构化稀疏;随机相位调制;非线性重构算法
英文摘要: Compressed imaging is one of the most core applications in compressed sensing theory, compared to classical imaging mode, which has more potential advantage in cost, portability, power, and high-resolution etc, however, existing compressed imaging methods often have defects in the long exposure time, or the high implementation cost, or lots of calibration work required. In previous research, we have completed some first stage work around the design of measurement matrices in compressed imaging, which demonstrate the feasibility of compressed imaging using random phase modulation, and the effectiveness of the modulation with low cost and deterministic phase mask. .The study intends to further exploit the basis of preliminary studies and investigate a compressed imaging model of the complex optical field through random phase modulation in the frequency domain under natural lighting conditions, which bridges the relationship between the complex optical field and intensity that can be directly recorded by optoelectronic devices..Meanwhile, based on Fresnelets which is most suitable for diffracted light field and the special structure in complex optical field, a more compact structured sparse model is constructed to investigate the effects of physical constraints of natural scenes and structured sparsity model on the reconstruction algorithm. For the amplitude and phase of the complex optical field coupling between each other, by improving the classical reconstruct algorithm to accurately reconstruct the original object from the real measurement data. The research of the project can provide some theoretical, computing, and technical support for the new generation imaging design.
英文关键词: Compressed Sensing;Compressed Imaging;Structured Sparsity;Random Phase Modulation;Nonlinear Reconstruct Algorithm