1:16.000 ölçekli hava fotoğraflarından SYM üretimi ve doğruluk modellemesi

DEM (Digital elevation model) production from 1:16.000 scale aerial photographs and accuracy modeling

Özet

Hava fotoğrafı ve uydu görüntülerindeki veriler, çok uzun zamandır klasik yollarla ve operatörler tarafından elle çizilmekteydi. Bilgisayar teknolojisi ve dijital görüntü işleme alanlarındaki gelişmeler, günümüzde bu işlemlerin otomatikleşmesine olanak sağlamaktadır. Otomatikleşmenin hedefi hızı arttırmak ve değerlendirme masraflarını azaltmaktır. Bu kapsamda fotogrametri alanında otomatikleşmenin büyük oranda sağlandığı işlemlerden birisi de Sayısal Yükseklik Modeli (SYM) üretimi olmuştur. Bununla birlikte otomatik yöntemlerin doğruluklarını artırmak amacıyla birçok çalışma yapılmasına rağmen, birçok alanda klasik yöntemlerin doğruluklarına ulaşılamamıştır. SYM üretiminde klasik yöntem olan fotogrametrik değerlendirme ile eş yükseklik eğrilerinin çizimi ve otomatik yöntem olan görüntü eşleme ile SYM üretimi ve doğruluklarının araştırılması konusunda çeşitli çalışmalar yapılmıştır. Ancak bu çalışmalarda elde edilen sonuçlar kullanılan kaynak veriye, üretim yöntemine, topoğrafik özelliklere, üretilen SYM’nin çözünürlüğüne ve karşılaştırmada kullanılan referans veriye göre farklılık göstermektedir. Bu çalışmada iki farklı bölgede (Bursa ve Ankara) 1:16.000 ölçekli hava fotoğrafları kullanılarak fotogrametrik değerlendirme ile üretilen eş yükselti eğrilerinden ve otomatik görüntü eşleme ile SYM üretimi yapılmış ve her iki yöntem doğrulukları bakımından karşılaştırılmıştır. Çalışmanın ikinci aşamasında Bursa bölgesinde üretilen SYM kullanılarak SYM doğruluğu eğim ve çözünürlüğe göre modellenmiştir. Model olarak birinci ve ikinci dereceden bir polinom seçilerek, polinom katsayıları dengeleme ile hesaplanmıştır. Elde edilen sonuçlar, SYM doğruluğu ve çözünürlüğün büyük oranda korelasyonlu olduğunu, SYM çözünürlüğündeki düşüşle beraber SYM doğruluğunda da düşüş gerçekleştiğini göstermiştir.

Abstract

Aerial photographs and satellite imagery have been evaluated manually by the operators for a long time for the extraction of the vector data. Computer technology and digital image processing technologies have been developed and this development provides to perform these extraction processes automatically or semi-automatically. The aim of making the processes automatic is to increase the speed of collecting the data and to reduce the cost. Digital elevation model production has been one of the most automated process of photogrammetry. Although, formats and accuracy for some formats of digital elevation models are standardized, data sources, quality of data sources and production methods of digital elevation models to provide or improve the standards are not determined clearly. Nonexistence of standards and production criteria causes loss of time and money and production of digital elevation models without known standards and accuracies. Also these nonstandard productions cause data loss of digital elevation models which are needed dense work and source. Many researches are made about the production of digital elevation models from different data sources and with different methods. In all of the studies, digital elevation models having different resolutions and data sources are produced and the accuracies of digital elevation models are investigated. The results of these studies vary according to the source data, resolution, ground control point distribution, production method, topography and reference data for the accuracy assessment. On the other hand, the accuracies are investigated in these studies, but the duration of digital elevation model productions are neglected generally. This situation prevents the optimization of production. Photogrammetry provides the most frequently used data sources and techniques for generating digital elevation models (Stocks & Heywood, 1994), either by direct generation of digital elevation models or indirectly via its use in topographic mapping for production of contour lines. Photogrammetry either involves stereoscopic techniques for interpretation of aerial photography or digital image correlation applied to aerial photographs. In this study, digital elevation models are produced by automatic image matching from 1:16.000 scale aerial photographs and the accuracies of digital elevation models are assessed. Also digital elevation models are produced from contour lines compiled from the same sources. Optimum data collection interval and accuracy of digital elevation models are investigated. Automatic image matching and photogrammetric compilation methods for digital elevation model production are compared. Accuracies are also investigated according to the resolution of digital elevation model and slope of topography. Some solutions are suggested for the areas having high errors. Two different test regions are used for the study. First region is Gölbaşı near Ankara. Gölbaşı is a test region used for different purposes for aerial photography. 40 1:16.000 scale aerial photographs of the region were taken in 2002. Second region is an area at the proximity of Bursa. 1:16.000 scale aerial photographs of the region were taken right after the 1999 Gölcük Earthquake. Five meter interval contour lines were compiled from stereo models for Gölbaşı region. First digital elevation model was produced from these contour lines. At the second stage, a digital elevation model was produced from the same area by automatic image matching. These digital elevation models are compared with 27 check points which are measured at the field by GPS. Root Mean Square Error (RMSE) of the digital elevation model produced from contour lines is found to be 0.61 meter. RMSE of digital elevation model produced by automatic image matching is found to be 2.98 meters. Five meter interval contour lines were compiled from stereo models for Bursa region also. Nine digital elevation models with the resolutions changing from 1 meter to 256 meters were produced from the contour lines. Digital elevation models are compared with 25 meter interval main contour lines which were compiled more precisely. RMSE of the digital elevation model is found to be 0.61 meter. Also the accuracies of digital elevation models changing with the resolution of digital elevation model and slope of topography are tried to be modeled with a polynomial equation. The results showed that the accuracy is correlated highly with digital elevation model resolution.