SISTEM PENGGABUNGAN FOTO UDARA DENGAN ALGORITMA SIFT (SCALE INVARIANT FEATURE TRANSFORM); SYSTEM OF STITCHING AERIAL PHOTOGRAPH BASED ON SIFT ALGORITHM

Arumsari Diwangkaton, Sumiharto

2014 | Skripsi | PROGRAM STUDI ELEKTRONIKA DAN INSTRUMENTASI

System of aerial photographs with SIFT algorithm which designed to combine the input image of aerial photographs. This image was generated by fixed-wing UAVs. The output of system is display in the form of aerial photos broader. The system is designed to stitch the aerial photographs using SIFT algorithm, FlANN, RANSAC, and warpPerspective in OpenCv library. SIFT algorithm is used as a keypoint detector of input image. FLANN method to found the keypoint matching. RANSAC homography matrix is used for the search the good matches. WarpPerspective method is used for merging the two inputs which have a samiliar keypoint. In this study case, include testing of rotational variation, variation in scale and minimal variation intersection. Whatever the rotation variations tested, both still images can be combined. Maximum scale obtained from testing the scale variation 340x359pixel first input image is equal to 310%, whereas in the first input image 800x600pixel obtained for 480%. Minimum scale obtained in testing with a scale variations 340x359pixel first input image is 61%, whereas in the first input image 800x600pixel gained by 45%. Intersection minimal variation that can produce the output of photo merging is 6% and the minimum number of keypoints accords between the two images that can be combined is 5.

Kata Kunci : Aerial Photograph, Stitching Images, SIFT, FLANN, RANSAC