Image mosaicing with unmanned aerial vehicles in indoor and outdoor environments

Image mosaicing is the process of merging two or multiple images with each other, making use of overlapping areas. Image mosaic is not used with unmanned aerial vehicles only. Generally, it is used in applications where the displayed area is small, but the viewing area is large. In this thesis, images taken by unmanned aerial vehicles were studied on computer. There are a lot of studies in the literature about obtaining panoramic images. However, studies on image mosaicing in aerial images taken by unmanned aerial vehicles are less. In the thesis, the image mosaic process is explained step by step by giving images. In the study, the feature extraction step which is one of the image mosaic steps is emphasized. Feature extraction algorithms in the literature were examined. According to the types of aerial images for the image mosaicing accuracy and performance in terms of appropriate feature extraction methods were tried to determine. Within the scope of the thesis, various applications have been made on indoor and outdoor aerial images. The accuracy and performance of the mosaic outputs were compared with different feature extraction algorithms of aerial images obtained from sloping ground and flat ground at different overlap values. Image mosaicing performance difference between two and multiple input images was examined. The image mosaic applications performed in the open source development environment and the applications performed in the non-open source development environment were compared. The indoor air images in different image sets were mosaiced with different feature extraction algorithms and the accuracy of the mosaic outputs was examined. In addition, the fact that the air images are taken from different viewpoint and in different illumination situations causes seams on the resulting image. In the thesis study, it was applied on the removing of these seams.