【作者】 杨丽；

【导师】 周颜军；

【作者基本信息】 东北师范大学 ， 计算机软件与理论， 2010， 硕士

【摘要】 图像匹配是指把两个或多个不同传感器所获取的同一地区的景物图像进行比较,确定出来或找到它们之间对应关系,以确定两幅图像之间的相对位置的过程,是一种重要的图像分析与处理技术。图像匹配技术是数字图像处理领域中一个重要的研究方向,近年来图像匹配技术被广泛应用于计算机视觉、资源分析、医学图像配准、光学和雷达跟踪、飞行器巡航制导、导弹地形匹配及投射系统的目标制导等。随着数码产品的普及,数码相机正在逐渐成为人们快速获取图像和计算机图像信息输入的常用设备。然而,数码相机的动态范围很有限,只能记录有限范围的对比度,亮度和颜色数,很难做到全面记录光的信息。而实际生活中,当拍摄场景的动态范围超出了照相机的采集范围时,仅仅通过改变相机曝光度来控制所需捕捉的亮度信息范围,成像总会有曝光过度或者曝光不足的区域,往往造成高亮处或阴暗处细节的损失,难以实现完整场景的再现。一组同一场景不同曝光度的图像可以比单幅图像提供更多的信息,较暗的图像可以提供场景明亮处的细节,而较亮的图像可以提供阴影处的细节。利用一组不同曝光的图像,可以融合处理成为一张光照动态范围较广的合成图像,但是首先一步是需要获取一组匹配良好的图像。本文是在经典SIFT算法的基础上,提出了改进,通过建立亮度变化空间,在亮度变化空间上提取尺度不变特征点,进行特征描述,特征匹配。通过实验可以得出,改进后的方法在亮度变化、尺度变化、旋转变化同时存在较大变化时,仍能取得较好的匹配效果。更多还原

【Abstract】 Image matching,is a process to match two or more images of the same scene taken by different sensors, finding the correspondence between them,and it is an important image analysis and processing technique. Image matching technology is an important research direction in the field of digital image processing ,being widely used such as computer vision, resource analysis, medical image registration, optical and radar tracking, aircraft guided.With the popularity of digital products, digital cameras are gradually becoming a common device in the real life.However, the dynamic range of digital camera is very limited, and only record a range of contrast, brightness and colors, it is difficult to record a comprehensive optical information. When the filming scene dynamic range exceeds the scope of the camera collection, it usually controls the range of brightness information which is achieved by changing the time for exposure. But no matter how to adjust the time, that always still exist overexposed or underexposed areas which leads to some detail losses of highlighting or the darkness department.A group of images with different exposure but same scene can provide more information than a single image. Darker images can provide some details of bright scene, and lighter images can provide the details of the shadow scene. The necessary task is to obtain a series of matching images. In this paper, a method for matching multi-exposure images is introduced. By extracting feature points from the brightness-spaces which are built by multiplying the contrast stretching function with a series of parameters by the input image, we can obtain interesting points robustly even under large illumination, rotation and scale changes. We use the scale invariant feature transform (SIFT) description to describe these points. The experimental results show that the proposed algorithm has good effects on dealing with the matching of multi-exposure images.更多还原