Appendix VIII

Automatic Mapping

Aerial photographs for mapping are obtained from an airplane that is in straight and level flight, the photographs being exposed at intervals of about six-tenths of the altitude. The camera commonly has a six-inch focal length and produces a nine-by-nine inch format film negative. The individual pictures are printed as positive images on glass so that they don't change dimensions with time.

The Automatic Stereomapper

Adjacent pictures taken from one flight line have a substantial overlap that permits stereoscopic viewing. Most commonly, they are mounted in a device like a Kelsh Plotter, as described in the text. For the Automatic Stereomapper, a Kelsh Plotter was used and set up in the usual manner to obtain the stereoscopic image.

Fig. 8-1 Kelsh Plotter Implementation of Automatic Stereomapper

Fig. 8-1 (left) shows the optical projectors and a portion of the model surface created by them. The viewing screen is replaced by a Nipkow Disc, shown separately in Fig. 81 (right). It consisted of a circular disc with holes laid out at equal angles but on a constantly changing radius. As the disc rotated in front of a rectangular window, the holes moved across the window, so the light through the holes were converted by the photocells below to corresponding TV-like electrical signals.

The diagram illustrates the situation where the Scanner is below the model surface as defined by the coincidence of rays from a bright point P in the scene. For this situation, with a hole in the disc moving to the left, the image from projector A appears earlier than the image from projector B; this provides the needed information to control the height of the Nipkow Disc so that it moves up to the point of coincidence of the rays. The mechanization of this is described next.

Fig. 8-2 shows idealized electrical signals corresponding to the point P. With the scan moving from right to left, the signal from A is seen first followed by that from B. The signals are assumed to have a duration W and unit height, so their areas are W.

Fig. 8-2 Output of Two Projector-Photocell Combinations.

In the height-error determining circuitry, these signals are passed through a multiplier to determine their degree of overlap, the product being their common area. In the position shown in Fig. 8-2, there is no overlap, so the multiplication would yield zero, but there would generally be some diffuse images is the film that would start the scan upwards. The signal from B is shown offset by an amount D. As the scanner moves upwards towards the correct height signals A and B move together until they touch when D=W; they then move to a complete overlap where the product equals their common area W. Further upwards motion of the scanner causes them to separate; the product again goes to zero at D=-W where the two rectangles just touch before separating. As shown in Fig. 8-3 there is a maximum output W when there is no error and the output falls off as the scanner moves from the correct height in either direction.

Fig. 8-3 Multiplier Output Showing Peak When Signals Coincide.

With the scanner at some height the product obtained is a single value that provides no information with respect to the error; it would take a trial and error procedure to determine the height of maximum signal from such information.

Fig. 8-4 shows the method used to convert the product to an effective error signal.

Fig. 8-4 Height-Error Sensing Unit.

The height-error sensing unit includes two multipliers and associated delay units and an element that subtracts one multiplier output from the other to yield the error signal. The top multiplier has a delay in the A signal link while the lower multiplier has the delay in the B link.

Fig. 8-5 shows the outputs of the two multipliers with the added output shown as positive and the subtracted output as negative. The two delays are assumed to be W/2, so there is a relative delay of W between the two outputs making the downward slopes of the two overlap to yield an error function that is positive for large negative errors, proceeds downwards with double slope through the zero error point and then goes negative. The central region, where the error is positive for negative errors and negative for positive errors is appropriate to drive the motors to change the height of the scanner -- for negative errors the signal is positive, so the height motor would drive to increase the height of the scanner until the error became zero.

Fig. 8-5 Difference Signal as Function of Height Error.

The actual signals were, of course, much more complex than the rectangular pulses shown, but the operation proceeded nearly as described. However, the customer had thought that in areas of steep slope in the direction of the scan the scanning window should conform to the slope, so the height error on the two sides of the scan center were measured separately and used to control a second motor that controlled the tilt of the scanner. This increased the mechanical and electronic complexity of the unit and was found later to be of little value in making the scanner follow the surface.

The UNAMACE

In the UNAMACE separate precision tables were used to hold the diapositives, so there was no real stereo model as set up by in the Kelsh Plotter. Instead, the associated computer directed the two tables to move such that the scanners were positioned on areas of the diapositives calculated, using an estimated altitude, to be corresponding images of the ground area at the geographic point of interest.

Fig. 8-6 Computer Controlled Diapositive Scanning.

Fig. 8-6 shows the basic scanning operation during stereo measurements. For the purpose of the explanation, it is assumed that the survey aircraft was traveling from left to right with A exposed first and then B, so the area of interest is at the right of diapositive A and at the left of diapositive B.

The scanning proceeds from left to right along a line parallel to the flight line and therefore along the line of parallax displacements associated with height errors, thus corresponding closely to the scan of the Nipkow-Disc of the Automatic Stereomapper except that the electronic scan permitted a much higher resolution and made possible operations with unusual forms of photography (the scan could be shaped like a parallelogram, by computer commands, to follow distortions in the photographs). The flexibility also made it feasible to use a large scan for operator viewing and a smaller scan when making the automatic height measurements; the large scan provided the operator valuable context information while the small scan allowed a more meaningful height measurement in sloped areas.

The height-error sensing circuitry used the multipliers and delay lines as shown in Fig. 8-4 for the Automatic Stereomapper, but in this case the height error was required to be in digital form. The Digital Height-Error Sensing Circuit is diagrammed in Fig. 8-7. It shows the height-error signal going to an accumulator whose output is the product of the error by the time. Assuming a positive height error, the output of the ACCUMULATOR would increase at a rate corresponding to the error, until it tripped the THRESHOLD DETECTOR. This resulted in a positive count to the REVERSIBLE COUNTER and the restoring of the zero-error condition at the output of the ACCUMULATOR. The count also caused the D/A converter to step the scan on one diapositive, by one height unit, in the direction required to correct the height error.

Fig. 8-7 Digital Height-Error Determining Circuitry.

The scan continued with the smaller error. If there was still some height error, the output of the ACCUMULATOR would again increase until the threshold was crossed, putting a second count in the REVERSIBLE COUNTER and a corresponding height correction into the scan. In steeply sloped areas, the count at the end of the scanning (usually 1/50 second, but longer in steeply sloped areas) could get as high as sixteen.)

At the end of the scanning operation, the number in the REVERSIBLE COUNTER was input to the computer, so it could correct its height estimate for the area, and the REVERSIBLE COUNTER and ACCUMULATOR were reset to zero. The computer then output the coordinates of the next point in the measuring sequence, using the updated altitude of the last measured point as the estimated altitude. The operation was then repeated on the new area.Proposed Modification of the Automatic Stereomapper.

The Stereomapper implementation was sluggish in operation because it was required to keep the mechanical scanner "on the ground". After work on the AMCS (the breadboard of the UNAMACE) had progressed it was realized that the operation of Stereomapper-type instruments could be greatly enhanced by the Digital Height-Error Determining Circuitry shown in Fig. 6-7, using the count in the REVERSIBLE COUNTER to step a delay line to compensate for the displacement in time between the two signals. This would make possible almost instantaneous corrections for the time errors associated with misadjustment of the height; the altitude reported at the output would then be made from a combination of the mechanical and digital values. This was never implemented but I did receive a patent on the idea 3. The Zeiss Corporation indicated an interest in this equipment but we never received an O.K. from the government to pursue talks with them.

Back to Automatic Mapping

Last revision: 3/9/97

Top of Page
Back to Table of Contents
Back to Sidney Bertram
Back to Jacques & Anne
Back to The Abramovich & Wilder Families
Back to Bertram Home Page
Master Index