Airphoto Seamless Mosaic Algorithm Without Ground Control Point
Dongli Fan · Yingxiang jishu · 2011
For aerial image mosaic,because airphoto splicing error accumulates in the aerial strips so as to cause the problem of later image distortion,in this paper a bundle adjustment technique was used to reduce error accumulation in the process of stitching.The algorithm used SIFT (scale invariant feature transform) to extract and match feature points;and Improved RANSAC (random sample consensus) purifies feature points,excluding outliers;then eligible points were used to calculate value of homography matrix between images by the least square method,on this basis,bundle adjustment method was used to entirely optimize homography matrix in order to solve image distortion in the period of post-splicing,thus mosaic image was accomplished.Finally,the dynamic weighted fusion method was used to better achieve seamless transition in the image seam.To verify the validity of the algorithm,splicing experiments using UAV aerial image sequence achieved good results.