Editorial

Abstract

Introduction

Site Details

Methodology
   Data Source
   Ground Survey
   Image Prep.
   Triangulation
   Digital Terrain Model
   Orthophoto manipulation

Results

Conclusions

Acknowledgements

References

Case Studies Index

3 Methodology

3.1 Data source

The scanned aerial photographs were transferred onto CD in the Helava Vitec Tile format (1:10000 scale) and standard Tiff format (1:2000 scale). These images were imported directly into the (DPW). The DPW runs a software package called Socet Set that is used to carry out all of the image manipulations and digital terrain model generation detailed in this study.

3.2 Ground Survey

Accurate control points are needed to tie the pair of stereo images to each other and to facilitate the construction of an accurately constrained product for spatial measurements. For the 1:10000 scale imagery Ground Control Points were identified from the aerial photographs and were surveyed in the field using a Leica Wild GPS 200 (differential GPS system). For the 1:2000 images, it was possible to place marker sheets in specific positions around the site that were included in the acquired images. The absolute positioning of these sheets was measured prior to image acquisition. GPS data was processed and geographic latitudes, longitudes and elevations generated using Leica SKI v 2.01 software. The values for each GCP were then input into the DPW to allow accurate calibration and rectification to be carried out on the images. In preparation for the more detailed examination of the stone circles, relative heights of selected stones were also recorded for comparison with the digital terrain models to be generated.
The main strength of the DPW method is that ground control can be constrained with a minimum number of measurements. For this study, eight points were collected in total that lie within the overlapping area of the stereo pair of aerial photographs. These points were obviously centered around the circle area to give the tightest fit for the DTM in this area. To give some idea of the relative sizes of the areas we are dealing with, and also to give some feel for the detail of the work being carried out, Figure 2 shows the original untreated images at 1:10000 and 1:2000 scales. Figure 2 shows the total area of coverage possible with 1:10000 imagery, and its application in topographic modelling as well as the finer detail possible at the 1:2000 scale for detailed spatial analysis.

Figure 2. 1:10,000 (A) and 1:2,000 (B) untreated scanned images of the CSC site and surrounding area. Stone circles in the 1:10,000 image are highlighted in a yellow circle

3.3 Image Preparation (orientation)

When images are obtained by photographic techniques, aberrations in the lens can distort the image. Clearly, these distortions give a significant error to any measurements made upon the images and prevent the accurate registration of the images that allows three-dimensional viewing. These are removed by the DPW by inserting the details of the lens distortions provided in the lens calibration report that accompanies the image files. Using the lens calibration files this first stage in image preparation was carried out. This procedure is referred to as Interior Orientation and is a basic procedure in image processing. The next step in image preparation (Exterior Orientation) includes two distinct processes (triangulation and rectification) that are used to prepare the images for stereoscopic examination and data extraction

3.4 Triangulation and Rectification

The images were registered to the ground using the Helava Automatic Triangulation System (HATS) component of Socet Set. Initially, the DPW creates a number of points on each image and finds the corresponding points in overlapping areas of both images (these are called Tie Points).
When this is completed, the latitudes, longitudes and heights for each of the surveyed ground control points are entered into the system and the position of each pinpointed on both images. The HATS module then highlights points where the system is not satisfied with the correlation between specified points on the images (expressed as a RMS statistical measure). Further processing of the image is carried out necessary until the system automatically locates all of the tie-points and ground-control points within threshold accuracy limits.
The images are rectified further to remove any small distortions in the images caused by slight changes in the flight path between capture of each image.
With interior orientation, triangulation and rectification complete, the stereo image is in a suitable condition to generate the digital terrain model and examine the site.