A parallel distributed system for SAR image processing
Abstract
Synthetic Aperture Radar (SAR) image focusing involves the processing of a large amount of data, thus requiring a very high elaboration time if a sequential machine is used. Should high precision be desired on the final image, the computer time can increase prohibitively. A high precision compression technique based on a two dimensional approach is described. This technique uses two stages two dimensional FFT (Fast Fourier Transfer) codes coupled with an asymptotic evaluation of the unit response function. An attractive feature of the above mentioned algorithm is that it is amenable to parallel processing due to the introduced decomposition of the FFT code. In SAR applications some processing systems capable of dramatically reducing computational time are presented. For real time applications VSLI (Very Large Scale Integration) dedicated systems can be used if some loss in image quality is acceptable. For high quality SAR processing the use of multicomputer systems is more appropriate, allowing a drastic reduction of processing time, while preserving the elaboration accuracy and flexibility. A parallel distributed architecture especially designed for the above mentioned SAR processing algorithm is presented and its global performance is evaluated. The design of a prototype machine IMAGO (Innovative Multiprocessor Algorithm for Geodetic Observation) is presented.
- Publication:
-
Applications of Multifrequency/Multipolarization SAR in View of X-EOS (X-SAR for EOS) CGS
- Pub Date:
- May 1990
- Bibcode:
- 1990apmm.symp..167F
- Keywords:
-
- Computer Systems Design;
- Data Processing;
- Earth Observations (From Space);
- Geodesy;
- Image Processing;
- Synthetic Aperture Radar;
- Algorithms;
- Architecture (Computers);
- Distributed Processing;
- Imaging Radar;
- Parallel Processing (Computers);
- Remote Sensors;
- Communications and Radar