06.22-P

The Role of the Image Processor in Planetary Instrument Definition

C.A. Robinson, F. El-Baz, K.C. Seto (Boston University)

Image processing techniques are utilized to evaluate various types of satellite data of the Earth, and to establish which are most suitable for correlation with certain features. This might help define payloads for future Mars' missions. For example, in consideration for fluvial feature identification and interpretation on Mars, a geomorphological examination of fluvial channels in the Eastern Sahara has been carried out. The Eastern Sahara is considered a suitable testbed for such studies because surface conditions are similar to those on Mars, including ample evidence for previous episodes of fluvial activity, with present-day domination by an active aeolian regime.

In order to assess the usefulness of different data types in feature recognition, a variety of image filtering techniques have been applied to SIR-C, Radarsat and Landsat TM and MSS data, for a region of southwestern Egypt selected for drainage mapping.

It is known that radar data can penetrate below the surface in the desert environment to highlight sub-surface riverbeds, whereas Landsat data portrays surface features. This suggests that radar data should be used to locate fluvial features on Mars covered by a thin veneer of dust. In this case, the spectral information would have to be forsaken. In order to test whether this need be the case, Landsat data were image processed, in areas where `radar rivers' had been identified, to see if these features may be unveiled by filtering. Preliminary results show that once the location of a channel is known from radar data, processing of the Landsat data can reveal the same drainage features, although to a lesser degree dependent on the depth of the sub-surface feature. Thus, in the absence of radar data, it may be possible to use Landsat data, processed according to an established set of image-processing routines, in this way. This work will be extended for other data types and data bands and using different processing techniques, such as, image classification.