Multisensor fusion algorithms for object detection, using subsurface data.

摘要

The objective of this work is to develop an automatic method by which the subsurface images which are data from the Ground Penetration Radar could be analyzed both in two and three dimensions and the objects or layers could be extracted from the image showing their exact positions.; The radar subsurface data contains background clutter which must be carefully removed for the object signatures to be recognized. Threshold filter with different threshold levels was used for these purposes. The Fourier Descriptors method was used for the description of the shapes of the objects. Classification methods based on artificial Neural Networks and Fuzzy Logic were implemented.; These methods were tested with Ground Penetration Radar data from the GSSI SIRveyor-20 equipment in the test field of UPRM, and data from NASA-SSC center. The results indicate that developed method can be very useful in subsurface object detection and depth determination.; Sensor data fusion is implemented using the data available (from a web site) from several sensors which scanned the foundation of a military camp. Several fusion methods are applied to this data and the results are analyzed, in this work.