【作者】 陈浙泊；

【导师】 陆祖康；

【作者基本信息】 浙江大学 ， 光学工程， 2009， 博士

【摘要】 人类获取的信息大部分来自于视觉,图像是视觉获取信息重要来源。数字成像技术在现代信息技术中有着举足轻重的作用。本文针对提高数字成像系统动态范围的方法进行了理论与实验研究。本文首次将信息论引入对数字成像系统动态范围的研究中来,明确揭示了图像的信息量与动态范围之间的关系:在系统曝光量不变的前提下,成像系统动态范围与图像信息量的负倒数成线性关系,即成像系统的动态范围越大,其相应获取图像的熵就越大。从信息论的角度系统的总结了目前扩展成像系统动态范围的三个方向:时间域扩展,空间域扩展以及自适应扩展。提出三种新的利用空间光调制器(SLM)实现的动态范围扩展方法:时间域SLM多次曝光,空间域光强编码调制,自适应PID反馈控制。完整的推导了三种扩展方法对动态范围的影响,并得出结论:在理想情况下,三种方法可以达到的最大系统动态范围为空间光调制器动态范围与图像传感器动态范围之积。通过数值模拟证实了这三种方法的有效性。自行设计并成功实现了基于硅基液晶调制器LCoS和CCD传感器的的高动态范围成像系统。利用该成像系统分别进行了时间域多次曝光,空间域光强编码调制,自适应PID反馈控制三种动态范围扩展实验。对每一种方法都给出了实验结果,并对实验结果进行讨论分析,最后对三种方法的适用范围进行了讨论。首次提出通过系统信息量计算来优化包含多离散器件成像系统的方法。该方法以获得成像系统最大互信息量为目标,寻找出各离散器件固有周期之间最优化的匹配关系,以及离散器件间最优化的空间位置,从而简单有效的削弱系统中莫尔条纹带来的影响。首次提出了一种基于不完全beta函数对高动态范围图像进行图像色调压缩的色调映射算法。该方法结合了全域法与局域法两种色调映射算法的特点,具有效率高,自适应好等优点,通过该算法可以有效的将高动态范围图像压缩到普通范围内显示。更多还原

【Abstract】 For human being, most of the information is acquired by human vision, and image is the most important visual information source. The digital imaging technology plays great pole in the modern information fields. Theoretical and experimental researches for the high dynamic range digital imaging system are carried out in this paper.Information theories are imported into the dynamic range research of imaging system for the first time. The relationship between information rate and dynamic range is revealed clearly: with constant exposure, system dynamic range has a negative reciprocal relation with the image entropy, which means the bigger dynamic range leads to bigger image entropy.According to the information theory view, three directions to extend the system dynamic range are concluded: temporal extending, spatial extending and self-adaptive extending. Three new dynamic range extend methods with spatial light modulator (SLM) are proposed: temporal SLM multi-exposure method, spatial light encoding method and self-adaptive PID feedback control method. The influence to the dynamic range of these methods are discussed, the theoretical maximum dynamic range is the product of the SLM dynamic range and image sensor dynamic range. Results of data simulation prove that the system dynamic range can be extended effectively.Designed and successfully realized a high dynamic range imaging system based on LCoS and CCD. Three dynamic range extending experiments are performed with this system: temporal SLM multi-exposure method, spatial light encoding method and self-adaptive PID feedback control method. Experimental results and discussion for each method are provide, the applicable range of each method is also discussed.Optimize method based on information rate calculation for multi-discrete-devices imaging system is proposed for the first time. By this method, the information rate of the whole system is taken as the indicator to seek the optimized matching relation among discrete devices, and also to determine the optimized position of these devices, and thus the influence of Moire fringe in the system can be reduced effectively.A new tune mapping arithmetic based on incomplete beta function is proposed for the first time. This method combined the global method and local method, and has high efficiency and good adaption. High dynamic range can be compressed into normal range to be displayed on general monitor.更多还原