ABSTRACT
The accuracy of the height determination is dependent on the base to height ratio, B/H, and the accuracy with which the parallax can be measured from the stereo pair. The National Map Accuracy Standards (NMAS) of the United States has defined the maximum allowable root mean square error in height [RMSE (z)] and plan [RMSE (x,y)] measurements, at a given map scale (Table 9.1). As seen from Table 9.1, RMSE (z), which is the total error in the height measurement of points in image data, dictates the contour interval applicable to a particular map scale. The planimetric error and the height error derived from stereo pairs are dependent on the pixel size. From Table 9.1, one can infer that if one can achieve a planimetric accuracy of ±1.5 pixel then for the 1:25,000 scale map, we require a spatial resolution of 5 m. Therefore, one can appreciate the need for high-resolution images for preparing large-scale topographic maps. Figure 9.2 gives, for a spherical Earth, the error in stereoscopic height δ h determination, as a function of B/H for different values of parallax measurement error, δ P (Wells 1990). As can be
seen from the graph, if the differential parallax can be measured with accuracy better than one pixel the improvement in accuracy of height determination is marginal as the base to height ratio is increased beyond about 0.6.
