期刊名称：International Journal of Computer Science and Network
印刷版ISSN：2277-5420
出版年度：2017
卷号：6
期号：3
页码：313-319
出版社：IJCSN publisher
摘要：Image mosaicking or image stitching is a process of combining two or more images to create a large panoramic image.
The resultant image can also be used for texture mapping of a 3D environment. Consider the case of images taken from a normal
camera. We can convert those images in to mosaicked image using mosaicking technique. This paper presents a new method of image
mosaicking based on SURF and VORONOI. SURF algorithm is used to extract features from the input images. Then the input images
are subdivided in to regions based on VORONOI method. In order to match the regions from VORONOI, ZNCC (Zero Mean
Normalized Cross Correlation) method is used and geometric transformation of input images are calculated based on that value the
two images are merged together to form mosaicked image. Finally to create a natural-looking mosaics, a quasi homography warp is
applied, which balances the perspective distortion against projective distortion in non-overlapping region. This method helps to reduce
projection time and execution time. It is faster than traditional mosaicking algorithms.
关键词：Image mosaicking; feature extraction; direct methods; SURF; VORONOI; ZNCC;
