【作者】 张媛；

【导师】 张铃；

【作者基本信息】 安徽大学 ， 计算机应用技术， 2005， 硕士

【摘要】 本文主要研究的是数字图像处理中常见的也是较难的一类:运动模糊图像的复原。在这里,图像复原是指去除或减轻在获取数字图像过程中发生的图像质量下降(退化),而我们的目的则是对匀速运动所造成的模糊图像进行处理,使它们趋向于理想的无模糊的清晰图像。 首先讨论水平方向的匀速运动模糊,介绍了多种复原水平匀速运动模糊图像的方法,针对退化模型已知的情况应用了魏纳滤波法来恢复图像,并且进一步介绍了Hopfield神经网络等方法分析了水平运动模糊图像的退化及恢复过程。 其次探讨了任意方向的匀速直线运动模糊图像的恢复,指出了这个恢复过程由三部分所组成:判定运动模糊图像的方向;评定运动模糊的参数;结合具体算法进行恢复。通过水平匀速运动模糊试验指出了现有的清晰度评价函数所存在的不足。并运用任意方向运动模糊鉴别尺度的新方法来恢复模糊图像。 本文主要是针对匀速运动模糊图像的恢复,日常生活中还存在很多非匀速运动的情况,这些情况可看成是匀速直线运动合成的结果,有待于我们进一步去研究。更多还原

【Abstract】 This dissertation primarily discusses a familiar and difficult type in image processing: restoration of motion blurred image. Here image restoration is that it wipes off or lightens fall of image quality in the process of getting digital image. Our intention is to deal with motion blurred image which is uniform, then to make it to ideal image which has no degradation.Firstly, The dissertation introduces motion blurred image which moves in horizontal direction. On the condition of knowing its degradation function model, this dissertation puts forwards several methods such as wiener filter, further introduces the process of motion-caused blurred image, establishing the mathematical model of degradation and restoration of uniform planar removing by using Hopfield neural network.Secondly, it discusses restoration of motion blurred image in arbitrary direction and points out that this restoration process is composed of three components: identification of the motion blurred direction; estimation of its parameter; restoration of a given method; then points out the shortage of existing clarity evaluation function by experimentation of motion blurred image in horizontal direction and uses a new method- identification the scale of the point spread function from the motion blurred image to restore blurred image.The dissertation mainly discusses uniform motion blurred image. In daily life there exists many motions whose motion speeds are not uniform . It can be seen as composition of uniform motion. We will look forward to research it in the future.更多还原