SISTEM PENTAUTAN FOTO UDARA DENGAN METODE LOCAL SYMMETRY DAN SIFT (SCALE-INVARIANT FEATURE TRANSFORM); AERIAL PHOTOS STITCHING SYSTEM BASED ON LOCAL SYMMETRY AND SIFT (SCALE-INVARIANT FEATURE TRANSFORM) METHOD

Ernawan, Dedhi F, R.Sumiharto

2015 | Skripsi | FMIPA

Aerial photography is usually used for mapping an area. The mapping process requires a picture with wide area coverage and clear so it needs aerial photos stitching system, one of the stitching process using the SIFT method. This method has good quality but relatively long on computation time. Local symmetry method is used to speed up the computing time that serves to reduce the amount of detected keypoint. The system uses SIFT detector in the detection process. Keypoints detection results then extracted using a local symmetry so that their numbers are declining. The system uses 2 kind of descriptor that are SIFT descriptors and local symmetry descriptor by calculating eigen values. The description Results further matched using FLANN. The next stage is the search pattern with RANSAC homography matrix and stitch the picture to accumulate keypoints using warpPerpective. Stitching system is tested with a few variations, that are translational slices variations, rotation variations, scale variations, and variations in brightness. Based on the test results, the method can reduce the number of local symmetry keypoints on all variations of the test with an average difference of 1513 keypoints. Computing time by using local descriptors symmetry is shorter than SIFT descriptors, the difference average between the two computing time is 226.93 ms. While the difference of average time between SIFT keypoints with local keypoints symmetry using SIFT descriptor is 641.67 ms. The results of calculation time stitching system overall shows a system using local symmetry and SIFT method had computation time shorter than the system using SIFT method with the difference average time 439.52 ms.

Kata Kunci : aerial photography; SIFT; local symmetry; photos stitching