Airborne InSAR DSM Anomaly Area Repair Research

Interferometric Synthetic Aperture Radar(InSAR)is an important technical means for obtaining three-dimensional surface information and deformation information.It provides information support services for natural disasters such as subsidence and landslides,and has great practical application potential.Is one of the research hotspots in the field of surveying and mapping.However,due to the unique imaging mode of SAR images,shadows and overlaps may occur in areas with large terrain fluctuations,which may cause correct phase unwrapping in some areas,resulting in low accuracy of the extracted Digital Surface Model(DSM).An invalid value has occurred,causing anomalies and vulnerabilities in the generated DSM.In response to this problem,this paper studies the detection and repair of the InSAR DSM anomaly area.The main contents and innovations include:(1)The current situation of the lack of feasible methods for the current InSAR elevation anomaly improvement,Combining the imaging geometry,deformation characteristics,product production process and InSAR elevation anomaly characteristics of SAR and InSAR,according to the principles of overall control and local adaptive processing,A complete set of InSAR DSM anomaly detection and repair solutions and implementation methods are proposed.It is possible to effectively find the elevation of the repair anomaly area under the premise of overall quality control,restore the shape of the abnormal area,and improve the elevation quality of the abnormal area.(2)InSAR data of different scenes in the survey area are different due to sensor acquisition position,viewing angle and time,so that there is high similarity between the overlapping images and there are differences in morphology and quality.In this paper,through the image area network adjustment,the InSAR orientation parameters of each scene are refined,and the geometric correction of the image,phase elevation and coherence coefficient map is performed by using the refined orientation parameters.Realize the geocoding of remote sensing targets and the geometric distortion of each scene data in the unified spatial reference of InSAR data of each scene.(3)According to the complex cause of InSAR elevation anomaly and the high redundancy of remote sensing target data,Comprehensive use of InSAR phase high data and interference quality information to achieve anomalous area detection of DSM,The DSM data with a certain overlapping area is processed according to the geocoded coordinates to obtain a difference map.The difference map is fused with the coherence coefficient map obtained by interfering with the DSM data,and the morphological processing is performed to obtain the DSM data anomaly and the void region.(4)For the characteristics of geometrical variation between the encoded overlapping data,The elevation repair method under the strict registration condition of local area is proposed.An image matching method based on improved SNCC algorithm is proposed.It is difficult to obtain a good matching effect by solving the conventional image matching method caused by large geometric deformation and lack of texture information in the mountainous region.Furthermore,the matching point pairs are used to establish the precise geometric transformation relationship between different DSM images in the same local area,and the overlapping data is used to repair the invalid values of the various elevations.(5)InSAR elevation hole areas that cannot be repaired for overlapping data,Patching by fitting and interpolation,The paper compares several commonly used interpolation methods and compares the results.The most accurate moving surface fitting method is used to carry out the interpolation test of the remaining cavity area.Complete the repair of the InSAR elevation.After repairing the DSM anomaly area,the paper spliced all the repaired DSM data in the survey area to generate a complete DSM product.The experimental results show that the proposed DSM anomaly region repair method can make the image contour change more natural,the terrain detail information is more clearly expressed,and the repair accuracy is improved,which indicates that the proposed method has good application value and application prospect.