摘要全景成像技术是一种通过为透镜阵列在非相干光源下就能实现的三维成像技术，这种三维成像技术是全真的三维成像。它能够产生具有连续视差的三维场景，并且不需要观看者佩戴眼镜、头盔等设备，是一种逐渐获得人们关注并且非常具有发展前景的立体显示技术。本文针对这种全景成像技术展开讨论，主要针对在实际拍摄过程中会出现的角度误差提出解决办法，研究的内容包含以下几个方面：68135

首先，为了获得模拟的记录图像，本文利用了光线追击软件POV-Ray来构造微透镜阵列模型以及被记录的三维场景，并且根据光学原理来构造成像系统。在此基础上，又模拟了当CCD记录阵列和微透镜阵列之间存在偏转角时的情况，从而为检验偏转角检测的方法提供模拟图。

其次，本文分析了在实际拍摄过程中可能出现误差的原因以及误差带来的影响。同时提出了基于Radon变换的网格线检测法，从而去检测网格线的偏转角并获得记录图像的偏转角度。

最后，根据实际情况中网格线往往不清晰从而不易被检测的特点，本文又提出了基于偏导的Radon变换法和结合边缘检测的Radon变换法用于提高网格线偏转角度检测的精确度，并通过构建加入噪声的模拟图像进行验证，最终得到了符合预期的结果。

毕业论文关键词  全景成像技术，微透镜记录阵列，Radon变换法

毕业设计说明书（论文）外文摘要

Title    On the Correction of the Images Captured by the  Microlens Array                                             

Abstract

Integral Imaging is a three-dimensional imaging technology which haves the lens array as its core optical device and can be achieved under non-coherent light. It is a kind of true 3D technology. It can generate continuous images and the viewers do not need to wear special observation equipment which makes it one of the most promising true auto stereoscopic three-dimensional display technology. Based on the introduction about this technology, this paper focuses on the correction of the images and includes following aspects:

Firstly, to obtain the recorded images, we use the Ray-tracer software POV-Ray to simulate a model of microlens array and a 3D view. At the same time we should build the imaging system according to the optical principles. By using this system, we can simulate the deflecting angle of the CCD and the recording array which could be used as the simulated image to test the arithmetic.

Secondly, this paper analyses the reasons that may course the errors in the real shooting process. Due to the angle errors, we propose the pre-processing method of captured integral image based on Radon Transform algorithm which can be used to detect the deflection angle of the grid lines and access to deflection angle of the recorded images.

Finally, because of the problem that the grid lines in the real captured image are much more difficult to be detected, we propose the some improvements based on the Radon Transform algorithm. One is about the partial derivative of Radon Transform and the other one is the combination of the Radon Transform and the edge detection method. To verified the improved methods, we add radon noise to the simulated images and use them to test the new processing method. 

Keywords: Integral Imaging; Microlens array; Radon transform

目   次

1 绪论1

1.1 三维显示技术简介1

1.2 全景图像技术1

1.2.1 全景成像技术的发展过程1