- Erciyes Üniversitesi Fen Bilimleri Enstitüsü Dergisi
- Vol: 38 Issue: 2
- Evolutionary Image Resizing based Accuracy Optimization for Aerial Triangulation
Evolutionary Image Resizing based Accuracy Optimization for Aerial Triangulation
Authors : Hacı Mustafa Palancioglu
Pages : 201-217
View : 15 | Download : 7
Publication Date : 2022-08-23
Article Type : Research
Abstract :Flight altitude and properties of the imaging-sensor influences spatial-resolution of aerial-images used for SfM photogrammetry. Most of the mini-UAVs developed for civilian use can provide 8-bit, 12-64 MP RGB aerial images with GNSS-geotags. As the number of aerial images obtained from related UAVs increases, various computational difficulties are encountered due to practical reasons for photogrammetric aerial triangulation process. Image Resizing is a common operation in SfM Photogrammetry for downsampling the aerial images. By using Image Resizing methods, images that are downsampled to different levels can make it easier to produce data that meets the desired accuracy criteria from aerial mapping. In order not to cause excessive loss of detail in the image produced at the end of the Image Resizing operation, the high-frequency data is partially suppressed by softening the relevant image before it is reduced. Over-smoothing the image or over-suppressing high-frequency data may cause altered, distorted, or lost details. Therefore, it is difficult to implement the Image Resizing operation in an ideal way. In this paper, Bilateral Filter is used to smooth the related aerial images for Image Resizing. The best values of the internal parameters of the respective Bilateral Filter were calculated using the Differential Evolution Algorithm, DE. The statistical spatial-accuracy values computed for the photogrammetric products obtained by using resized images with DE-based Bilateral Filter exposed that the related products can met the spatial-accuracy standards for various engineering applications.Keywords : Aerial-imaging, Multiscale DTM, Image Downsampling, Spatial Accuracy, SfM Photogrammetry