DSTO Publications Online >
DSTO Publications Online Repository >
DSTO Formal Reports >
Please use this identifier to cite or link to this item:
|Title: ||Functions for processing synthetic aperture radar imagery|
|Report number: ||DSTO-TR-0146|
|AR number: ||AR-009-210|
|Report type: ||Technical Report|
|Authors: ||Whatmough, R.|
|Issue Date: ||1995-01|
|Division: ||Information Technology Division|
|Release authority: ||Chief, Information Technology Division|
|Release limitations: ||PUBLIC RELEASE|
|Announce limitations: ||Public Announce|
|Task sponsor: ||DEF|
|Task number: ||93/163|
|Pages or format: ||44|
|DSTORL/DEFTEST terms: ||Synthetic aperture radar|
|Other descriptors: ||Ingara|
|Abstract: ||The INGARA synthetic aperture radar system can potentially benefit from higher level algorithms and software to enhance its basic imagery so that important features are obvious to human operators. These could include filters for removing speckle, zooming functions for correcting the aspect ratio of the imagery and classifiers for distinguishing ground cover types. The system could also make use of a development environment in which such functions can be easily developed and trialled. This Note describes the Khoros image processing system and functions added to that system by researchers in ITD that could be of use in processing INGARA data. It also includes results of tests of the effectiveness of these functions in enhancing sample INGARA data.|
|Executive summary: ||The INGARA synthetic aperture radar system developed by Microwave Radar Division is now producing images. These images will need to be further processed to make important detail visible to human operators. This report has been written to describe some of the image processing software and systems presently available to carry out such processing: it details appropriate algorithms and software currently held by Information Technology Division, and also presents an effective development environment for creating and testing new algorithms.
The Khoros image processing system has been developed at the University of New Mexico and is freely distributed over computer networks. It provides a wide selection of basic image processing functions and a graphical user interface through which these functions can be combined to build more complex operations. Users can develop their own specialist functions and add them to those provided.
Standard Khoros functions relevant to INGARA processing perform image display, several forms of filtering, grey-level manipulation, geometric rescaling, and the computation of local statistics for texture analysis. Information Management Group researchers have added functions that perform new forms of these operations, and also automatic image thresholding, supervised texture classification by artificial neural network and cleaning up errors in classifier output.
The above functions were tested on INGARA images, for their ability to alter the aspect ratios of images, to remove radar "speckle" without destroying other features of possible importance, to detect bright objects as "targets" of interest, and to classify surface regions using texture measurements. Various combinations and sequences of grey-level transformation, rescaling and filtering were tried. The output was inspected visually, and quantitative measures of speckle reduction and point target blurring were computed where relevant.
Not all functions were suitable for the data used. Various functions and combinations are recommended for geometric rescaling, speckle reduction, classification and target detection.|
|Appears in Collections:||DSTO Formal Reports|
Items in DSTO Publications Online are protected by copyright, with all rights reserved, unless otherwise indicated.