重轨极化干涉SAR无模糊目标分解及建筑目标提取OA

Polarimetric target decomposition and its application to man-made target extraction constitute an impor-tant topic in polarimetric synthetic aperture radar(PolSAR)image processing.However,conventional polarimetric de-composition tends to overestimate the volume-scattering component in oriented built-up areas,thereby degrading subse-quent building extraction performance.To address this issue,an unambiguous target decomposition method for repeat-pass polarimetric interferometric synthetic aperture radar(PolInSAR)is proposed and applied to building extrac-tion.First,the ambiguity-suppression capability of repeat-pass PolInSAR coherence is analyzed and its physical inter-pretation is discussed.Then,an urban descriptor is constructed based on coherence and incorporated into a target de-composition model,by which unambiguous polarimetric-interferometric SAR target decomposition is achieved.Finally,the decomposition results are integrated with multiple building extraction schemes,so that the performance of existing al-gorithms is improved and the effectiveness of the proposed method is validated.Comparative experiments are conducted on three PolInSAR datasets,including E-SAR,UAVSAR,and GF-3.The results demonstrate that,compared with con-ventional polarimetric decomposition methods,more accurate scattering features are obtained by the proposed method;when applied to building extraction in complex scenes,fewer false alarms and missed detections are produced.