Back Scattering Mechanisms Of Typical Terrain Cover Types And Its Application In Image Classification

Synthetic Aperture Radar (SAR) is an effective microwave sensor. It has the advantage of working day and night under any weather conditions. Images that acquired can be high resolution covering large areas. Frequency and polarization are two important features of electromagnetic wave. Targets behave different under different frequency and polarization. By polarimetry, additional information of targets can be gained besides magnitude and phase. Multi-frequency SAR is playing more and more significant role in the interpretation of terrain cover types.In this thesis, the back scattering mechanisms of typical terrain cover types and its application in image classification have been studied. The main work and contribution are as follows:1) The back scattering mechanisms of typical terrain cover types has been studied systematically. We made a deep investigation on several important parameters. The summary of the characteristics can be a reference for the classification.2) Large amount of references about target decomposition have been studied. Theories as well as the algorithms have been comprehended. Scattering mechanisms about targets have been exposed through these methods.3) Classification algorithm is also been studied and we made an improvement on the classification results by converting the research target from pixel to patch that can represent target's character. Based on six feature parameters, we used the decision tree classification method to classify the AIRSAR image and got a result with high accuracy.4) An effective software system for PolSAR image processing has been designed and developed.