7.2.3 Stereo Configuration

The two image pairs considered were F2/F1 (Stereo pair 12) and F4/F5 (Stereo pair 13) collected on descending/ascending passes respectively. At this latitude, (approximately 45º North), the convergence angle between ascending and descending orbital passes is between 20º and 25º. To compensate for this, the images must be rotated with respect to each other in order to obtain the stereo-model. These image pairs are very sensitive to mis-orientation. Once the images have been oriented correctly, the viewer will notice that compared to a same side image pair, a smaller image area is viewable in stereo.

Vertical exaggeration is pronounced in both image pairs (Stereo pair 12 and 13) due to large intersection angles. Thus, on these images, the topography around Sherbrooke appears to have much greater changes in elevation than the actual 250 m. The vertical exaggeration is more pronounced in the F2/F1 image pair since F1 and F2 viewing angles are steeper than F4 and F5 viewing angles. Stereo parallax is greater for steeper viewing angles. Elevation distortions increase with steepness of viewing angle.

The radiometry of all four images is very similar. There is a strong contrast on each image allowing lakes, rivers, roads, and even a ski hill to be identified. Forests and agricultural areas are easily distinguished due to the sensor's sensitivity to roughness. Seasonal changes in vegetation can be identified when comparing the F1, F2, F4 images with the F5 image. The F5 image was acquired in June1996, while the former three were all acquired in October, 1996. Soil moisture is difficult to identify without some knowledge of the land use/cover of the area.

The F2/F1 images display more foreshortening effects than the F4/F5 images. This is expected because the viewing angles of the former are steeper than the latter.

The first attempt at viewing these two image pairs (Stereo pair 12 and 13) may be confusing. The opposite side stereo configuration causes foreshortening and shadow effects to appear on opposite sides of land features in areas where slopes are greater than the viewing angles. Depending on the individual viewer, foreshortening cues may overpower shadow cues and brighter aspects dominate darker ones, or vice versa. This means that, at first, it will not be easy to see the depth of valleys or the heights of mountains in those local areas. However, with practice we can learn to be comfortable with these opposing effects. Depth information will be taken from the dominant cue. We can learn to see large differences in relief very well with an opposite side stereo configuration. Where the topography is rolling, stereo viewability is much easier. The viewer is encouraged to practice with the F2/F1, F4/F5 image pairs and to try the F2/F5 and F4/F1 combinations on their own.

The F1, F2, F4 and F5 used in the opposite side configuration combinations (Stereo pairs 12 and 13) can also be used to generate two same side stereo image pairs: 12A and 13A. F5/F1 (Stereo pair 12A) is from an ascending pass and F2/F4 (Stereo pair 13A) from a descending pass. The radiometric aspects of these image pairs are discussed on the Stereo pair 12 and on Stereo pair 13 pages.

These image pairs are easier to view in stereo than the opposite side combinations. A larger image area is visible (at any one time) in stereo since the images do not have to be rotated with respect to each other. The most obvious difference between these two same side image pairs and the opposite side combinations is that a same side stereo configuration does not produce the same amount of vertical exaggeration. Both of these image pairs display a much more realistic visualization of topography in the Sherbrooke area.

Due to larger intersection angles a stronger perception of relief is given in the F5/F1 image pair than in the F2/F4 pair.

This image pair is very easy to view in stereo. However, due to the relative steepness of the viewing angle, foreshortening and layover effects are noticeable. The ridge in the top half of the stereomodel appears to lean towards the sensor. Vertical exaggeration is pronounced. Standard mode viewing angles are steeper than Fine mode viewing angles so that vertical exaggeration in this pair is more evident than in the Fine mode pairs discussed in the previous pages.

This image pair is also easy to view in stereo. Since S7, the shallowest of the standard mode viewing angles, is used in this image pair, foreshortening and layover effects are less pronounced and shadow effects are stronger than in Stereo pair 14.

In addition to the S1 and S3 images which make up stereo pair 10, S2 and S7 images were also made available. These four images can be combined in order to generate stereo pairs 10A and 10B once they have been re-scaled to a common size. (Resizing is discussed in Part 4 of this chapter.) Stereo pair 10A was created from the S1 and S2 images. The radiometry of the Ottawa image pairs has been discussed previously. The most obvious feature of this image pair is that there is less vertical exaggeration than on the S1/S3 pair discussed previously and the S1/S7 stereo pair to be discussed next.

Stereo pair 10B was created from the S1 and S7 images. As with the S1/S3 and S1/S2 Ottawa image pairs, the Gatineau Hills and the Ottawa River valley are easy to see in relief. Relief on both sides of the river is more pronounced than in the S1/S3 and S1/S2 image pairs. Of all the same-side stereo Ottawa image pairs discussed thus far, this pair shows the greatest stereo effect. Vertical exaggeration is more pronounced in this image pair due to the greater intersection angle.