一种基于区域检测特征描述子的静态图像拼接算法

A STATIC IMAGE STITCHING ALGORITHM BASED ON LOCAL REGION DETECTION FEATURE DESCRIPTOR

图像拼接技术是当前图像处理领域的一个热门研究课题。它将一组相互存在重叠部分的图像进行空间匹配对准,经融合后成为一幅包含各图像信息的大视场的完整图像。提出了一种基于区域检测特征描述子的图像拼接算法。算法由以下几部分组成:首先,在图像配准方面,主要采用区域检测特征描述子来寻找特征点和特征描述子并据此对图像进行匹配;然后,利用统计学方法寻找高效匹配点;接下来,运用单应性矩阵投影方法对图像做投影变换;最后,基于变换结果的图像融合拼接。

Image stitching technology is a hot topic in image processing sector at present.Its job is to match and register in spatial a series of images with overlapped regions each other and fuse them into an entire image of panoramic view containing every image＇s information.In this article we raised an image stitching algorithm based on local area detection feature descriptor.The algorithm is consisted of four parts as follows： First,in image registration process,the local detection feature descriptor is used to search feature points and feature descriptors and the images are matched on that basis.Second,statistics method is employed to search efficient coincident points.Third,the projection transformation on the images is carried out using the homography matrix projecting method.Fourth,the images are fused and stitched based on the transformation results.