The position of one object (P) on 2 different photos is measured from the center of each photo (o1 and o2). This distance is called the radial parallax of that object. It depends on the position of the object in the images, but also on its heigth (a higher object generates a longer parallax). Measuring radial parallaxes is an indirect way to measure heigths.
In order to accurately measure radial parallaxes, one must know the position of the optical center of each camera (3 coordinates X,Y,Z ) and the orientation of the aiming axis of the camera (3 angles omega, phi, kappa). These 6 parameters define the external orientation of the camera (for each photo).
Cameras are not perfect. They generate geometric distorsions, summarized by several parameters describing the internal orientation. In normal conditions, these parameters are fixed for all photos acquired by the same camera; these parameters are found in the calibration certificate of the camera.
Using the images and internal / external orientation parameters, specialized software will automatically recognize many similar points in the overlapping zone between 2 photos.
These points are called tie points (green crosses on the following stereo pair).
Using these points in combination with orientation parameters (and some ground control points, if available), the software will define what was the position and orientation of the camera for each photo, and this will allow binding each photo to its neighbours, to create an homogenous block of images.
This process is known as "aerotriangulation".
The resulting blocks can be displayed in 3D, either as anaglyph or using a dedicated 3D display.
Photogrammetry software can then be used to measure the position and heigth of features, by moving a stereoscoipic cursor with adjustable radial parallax (X-shaped black cross on following examples)
Using this technique, one can produce complex 3D mapping of all objects visible on the aerial photos.
This technique was used to produce most existing topographic maps.