There are many problems in topographic mapping in an urban area. Traditional land survey is a very time consuming technique and can be very expensive. Photogrammetry is a popular choice but there are some problems such as clouds and limited operational time. Since Synthetic Aperture Radar, (SAR), is an active remote sensing system and its signal can penetrate through clouds, it can be operated at any time of day and is a independent of the weather. SAR could be a good solution for topographic mapping in an urban area. Combining SAR data and Interferometric radar technology can provide enough information for topographic mapping. Information can be extracted from SAR intensity Image. This thesis focuses on feature extraction and classification for topographic mapping in an urban area from airborne interferometric SAR data. A new algorithm is described which is simple and practical but yet very efficient for feature extraction and for object-based feature classification. An adapted Canny-Petrou-Kittler algorithm is applied for edge detection. Since the algorithm provides good detection, good localization and only one response to a single edge, it is an ideal edge detection for dense urban areas. Since the SAR image is noisy by its nature, small weak edges are expected. The modified non-maximal technique is also proposed to reduce unwanted edge. The technique of generation of bald earth DEM is proposed to obtain a normalised DEM for feature extraction. Region growing from edge detection is then applied to extract a more accurate shape of the feature and generate feature surface by using topographic parameters. The extracted feature is then classified by object-oriented classification technique, in which the classification is performed at object level not pixel level. And at the end of the process 3D city model can be produced.

In this thesis, we investigate the integration of local motion signals across space. We examined how both perceived speed and speed discrimination is affected by an increase in the number of local motion elements, their spatial separation, and their contrast. We found that the rate of integration was affected by contrast and the number of elements; the lower the contrast the greater the rate of integration. At a contrast close to threshold (7%) the rate of integration could be accounted for by probability summation of local speed signals across space. The effect of distance between speed cues was negligible. The improved speed discrimination sensitivity with an increase in element number could not be accounted for by increased detectability; for equally detectable stimuli, speed discrimination thresholds are decreasing with element number. Whereas element number strongly increases speed discrimination sensitivity, the effect on perceived speed is small and not always consistent across observers. Finally, the integrative process was found to be related to the direction of the local speed cue at high contrast (20%) whereas at low contrast (7%) the integrative process is direction blind. The implications of the results obtained within the context of this research are discussed.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.