Image processing via computer in aerial archaeology

Abstract

For more than half a century, aerial photographs made with hand-held cameras from light aircraft have made major contributions to the discovery of archaeological sites in northern Europe.' Almost magically, a picture taken from a high viewpoint makes the invisible visible. During this long period, enormous numbers of photographs made with every imaginable type of camera and film have preserved the only information on sites which have since been destroyed by modern construction. In the homeland of the method, Great Britain, there are surely over a million such pictures, and there are several hundred thousand others in some of the other countries of northern Europe. Despite this enormous accumulation of raw data, however, there has been very little in the way of synthesis of results. An earlier article pointed out the need for transferring the data to large-scale maps or plans.2 Why to maps? To begin with, the conditions which allow buried remains to be seen fiom the air are extremely complex. A common sign is discoloration in crops or soils. It is most unlikely, therefore, that all of a large site will ever be seen in a single photo taken at any one moment. But different details are revealed in pictures taken over many years under different weather conditions, crop development and agricultural treatment. A proper drawing on a large-scale plan can accumulate all of this information from year to year, and a set of such plans can display all the data relevant to a region compiled during decades. A second reason for creating maps is that identifiable ground features which allow one to locate sites may change with time as roads are displaced and new structures built. Hence it may be impossible to locate a site precisely without an absolute geographic reference frame. Only recording data accurately on large-scale plans enables an excavator to locate his cuttings accurately and not waste much time in sterile trenching. Since many of these sites are threatened with destruction through modern construction, it would be very useful if all data could be recorded on the 1:5000 base map, continuously updated. Copies of these maps with the archaeology clearly marked, placed in the hands of the planning authorities in our own ministry, could lead to the relocation of roads and buildings and prevent the high cost of excavation when sites are studied in the future. If there is no other way out, adequate warning of impending destruction can be obtained from the earliest planning phase. Then the sites can be excavated calmly without bulldozers working nearly on top of the archaeologists, as is often the case today. Since the problem of incorporating the older archives and the slowness of terminal entry of coordinate data from oblique photos measured by hand is evident, a new method was needed. What, then, is the problem? Why have not all photographs long since been recorded on the map base? The answers are simple: people have not thought it necessary, or suitable maps were not available, or they didn't quite know how to do it. A map scale of 1:5000 is needed for useful recording of details, and for positioning of an excavation, 1:1000 is required. In most countries of northern Europe maps at 1:5000 in heavily built-up areas where sites are most in danger are now becoming increasingly available. Still, although a number of simple techniques for transferring features from photos to maps have long been published in the photogrammetric literature, they appear to have been unknown to most archaeologists. The difficulty in transferring photo information from hand-held cameras to maps lies in the fundamentally different techniques used in archaeological air photography and in photogrammetric mapping. The pictures are made with the camera pointed obliquely at the site through an open window or