Features Extraction And Classification Research Of SAR Target Based On Morphology

With the development of the synthetic aperture radar (SAR) technology, theamount of information and the complexity of SAR images are increasing, and the needsof modern research are unable to be satisfied by the traditional methods. Obviously, theinterpretation capabilities of SAR image have become an important problem whichneeds to be solved urgently. Two important aspects of SAR image interpretation areterrain classification and extraction. Not only terrain classification is able to replace thetarget detection and identification module as a SAR image interpretation front endsystem to provide some position and scale information to identification process, but alsoprovide the necessary parameters directly to the manual interpretation to establish anindependent auxiliary interpretation system. Due to the impact of factors such as dataquality, the progress of SAR image terrain extraction is very slow. Until recent year,new data support is provided by a large number of high-resolution urban SAR imageand people began to concern about the subject, so that it became a new field of SARimage interpretation.The dissertation studies the SAR image interpretation and SAR image terrainextraction with the mathematical morphology theory. Main researches are as follows:1. SAR image classification theory and mathematical morphology basic theory andoperation are researched. Then we use mathematical morphology image processingmethod in SAR interpretation.2. The developed Morphological Profiles method based on morphological variablemetric method is studied. Then this method is used as the means of SAR terrainextraction and classification.3. Morphological Attribute Profiles method, based on expansion of MorphologicalProfiles method, which processed the image with attribute morphological operation withthe selected attribute, is researched. Then the results are used as features of terrainextraction and classification. The terrain extraction results are compared and theanalysis of improvement of the methods and the impact of the different attributes aregiven. 4. Mathematical Morphological Attribute Profiles are applied in SAR imagefeature extraction and terrain classification, and the analysis of the experiment is given.