7.0 Stereo Interpretation

The purpose of this chapter is to illustrate how a stereoscopic apparatus is used to view and interpret RADARSAT data in stereo. Data sets, from RADARSAT and other sources, will be used as examples to show:

• three different methods of representing depth information,
• possibilities of creating stereo image pairs from different data sources,
• different look directions, beam modes and beam positions of RADARSAT data available to the user,
• effects of ground cover, terrain, and geographical location on RADARSAT image pairs, and
• different image processing techniques that enhance the viewability of RADARSAT stereo pairs.

Stereoscopic viewing

The objective here is to build practical experience using theory provided in previous chapters of this manual with examples of RADARSAT and other data sets. The user will gain a practical understanding of the geometric and radiometric differences between different data sets. This will include the way in which parameters related to satellites, sensors, Earth location and image processing techniques are affected by those differences. Furthermore, information presented in this chapter will enable users to generate the best stereo pair according to the data set, study site and thematic application that they are currently working with.

Section 7.1 is a comparison of RADARSAT stereo image pairs with other platform and sensor image pairs. Data acquired by aerial photography, airborne SAR, SPOT and ERS-1 satellites will be compared with RADARSAT data in terms of geometry. In the case of sensors, image pairs obtained with a camera, VIR scanners and other SAR systems will be compared with RADARSAT images in terms of radiometry. Part 2 also contains examples of mixed sensor pairs such as SPOT/Landsat and ERS-1/JERS-1. These are included to demonstrate the possibilities of using data from different sources to generate depth perception. The geometry and radiometry of each mixed sensor pair will be compared.

In Section 7.2, stereo viewing of RADARSAT images will be addressed. Sensor parameters such as beam modes, beam positions, and look directions and ground considerations such as terrain type and geographical location will be considered.

Some effects of sensor parameters and ground considerations include:

• the amount of area that can be seen in stereo,
• the amount of detail that can be seen in stereo, and
• whether layover, foreshortening or shadowing present a barrier to effective stereo viewing.

To illustrate these factors RADARSAT stereo pairs, which are easily viewable in stereo will be compared with those, which are not.

In Section 7.3, problems that affect stereo viewability, but can be corrected by simple image processing techniques, will be discussed. Scale differences between image pairs, due to differing sizes of the data sets or to a slant range representation, can be corrected. Rotations can be done in order to allow the viewer to see an image pair in stereo with greater ease. An antenna pattern correction can be applied to an image where the antenna pattern interferes with the ability to see the image pair in stereo. Filtering techniques can be used to reduce speckle. Examples of image pairs, where these image processing tools have been used, will be discussed.

When viewing stereo image pairs it should be remembered that it will be easier to perceive depth in some pairs than in others. Viewing a stereo pair of aerial photos or SPOT satellite images will seem easy and "natural" because these are optical systems. Our eyes and consequently our brain perceive objects as optical systems do. Initially, SAR image pairs may be more difficult to view in stereo. This is due to humans not "seeing" as a SAR system does. We are not sensitive to microwave energy. Nor do we "illuminate" the objects we are looking at as bats and dolphins do using sound waves. Psychological cues, such as shade and shadow, which we rely on, are different on optical and SAR images.

It may take our brains a little time to integrate new information. Practice makes "perfect". With experience, SAR image pairs can be viewed in stereo as easily as aerial photographs. Additional experience enables us to view even mixed sensor pairs in stereo. People trained and experienced in stereo viewing may have a greater facility to perceive depth in challenging stereo pairs. Do not give up.