A Study on Automation of Image Mosaicking

• Main Authors: Ying-Ting Huang, 黃英婷
• Other Authors: Victor J. D. Tsai
• Format: Others
• Language: zh-TW
• Published: 2001
• Online Access: http://ndltd.ncl.edu.tw/handle/28134918587278541338

碩士 === 國立中興大學 === 土木工程學系 === 89 === Photogrammetry is widely used in change detection environment assessment, field investigations etc. However, it is difficult to cover the whole area of interesting with single photograph due to the limitation of the size of camera frame and the flying height of airborne. Therefore we have to mosaic all images that were taken continuously for following applications. Thus image mosaicking is a very important data process to the integration of images. This research investigates the processes on automation of image mosaicking, such as feature extraction, image registration, merge of images, and adjustment of image gray value, in order to reduce manually intervening. First, image feature points were extracted by using Lue interest operator and three pairs of control points were chosen manually to establish the coordinate transformations between both images to provide initial information of conjugate control points for enhancing the accuracy of image matching, and to extract seam line of both images. Before merging images, a look up table of the buffer zones of the seam line was established using histogram matching for the adjustment of gray values of the searching image, such that to the global contrast of the searching image is similar to that of the reference image for much smooth appearance of the mosaiked image.