Perspective figures may also be used. If the lines or subdivisions of the perspective figures are sufficiently close, detail may be copied by eye, correcting both for tilt and for scale in the process.
Photographic rectification is more suitable for larger areas if these are really flat. It must be remembered however that a sharp image is essential to good plotting and that the ordinary projecting lantern, though it offers freedom of movement for the screen, does not do so for the photograph. This latter move ment is, however, necessary to secure a sharp image. Space forbids a further analysis of the perspective and optical prin ciples involved. (See bibliography.) For small scale or exploratory mapping it may be convenient to assume that photography has been vertical and the height has been constant. Mosaics (com posite pictures made up of "vertical" photographs) are made on this assumption, although where good maps already exist, mosaics are usually built up on positions taken from them.
Examples of Surveys on the Above Lines: i. The Topographi cal Survey of Canada has made excellent quarter inch maps of the lake and river areas of the Laurentian Plateau from highly tilted oblique photographs in which the horizon appears, and serves the purpose of two of the four necessary control points.
ii. The swamps and forests of the Irrawaddy Delta were mapped by the Survey of India from "verticals" controlled by triangulation and traverse. Individual photographs were not rectified but strips of photographs were prepared and rephoto graphed to the correct scale as given by the control.
iii. The Air Survey of Flanders and Picardy in 1916-17 was based on a fresh triangulation, on which an old but fairly reliable cadastral survey was assembled. Fresh detail, inclusive of trenches and other military works, was added from vertical air photographs, each of which was treated as a perspective view within the narrow limits of the cadastral control.
(b) In Ordinary Country and for Contoured Maps.—In the bundle of rays which converge to the camera that from the top of a high chimney or hill (unless it be vertically under the camera) will make a larger angle with the vertical than will the ray from the point directly under that chimney or hill top. Heights are thus displaced outwards from the plumb point, depressions in wards. The only perspective effect which holds good is that of direction from the perspective centre.
In ground photo topography the position, height and tilt of the camera are known. In air photo topography none of these are known, but they can be found from the corresponding positions on ground and photograph of four points properly surveyed in plan and in height.
The reconstruction may be graphical, mathematical or me chanical. Graphical methods are not sufficiently precise, whilst a mathematical solution, possible in several ways, is excessively lengthy. The mechanical solution is that adopted in the machines
for stereoscopic measurement which will be mentioned later. All three suffer from the difficulty of securing suitable well defined and sharp images on the plate. Precision of measurement on the plate is therefore difficult to attain whilst the comparatively small area of the base (the area included by the four control points), and the comparatively sharp angles at the apex, are difficulties.
The stereoscopic machines for surveying from ground stations act on this general principle but the problem they solve is com paratively easy because the positions of the ground stations can be surveyed directly and the photographs can be taken at meas ured angles of tilt and in a definite angular relationship to each other. Air stations cannot be surveyed directly. Reference must be made from the images on the plates to the points of a ground control, which those images portray, before the stereoscopic image can be plotted on the plane of the map (e.g., mean sea level), properly oriented, and reduced or enlarged to the appro priate scale. This reference to ground control can be carried out in two ways. Firstly, as in those stereoscopic machines which follow the general lines of their predecessors designed for ground stereo plotting, each photograph is set, singly, upon a control of four points surveyed on the ground, thus reconstructing the posi tion and tilt of the plate in space. When both plates of the pair have been thus placed a final joint setting ensures proper corre spondence. The process of setting may take several hours. There is however another way which is to set the pair in correspondence, in space, before referring to the ground control. The pair is said to be in correspondence when corresponding directions in space intersect, each pair of points be ing in correspondence when the rays to them intersect. A per fect correspondence between plates is secured when five points are in correspondence and the plates are then set in their cor rect relative positions to each other and to the base line be tween the two air stations. This is the principle of the Fourcade stereogoniometer. The strip so set or plotted is, however, not on the required plane nor is it necessarily correct in scale and orientation. To secure correct plan the whole must be referred to a ground control which, how ever, may be comparatively open. We must know— (a) Scale—This is deter mined from a measured base on the ground by comparison with the corresponding length in the stereoscopic image.