Scene classification method based on interferometric synthetic aperture radar images

Abstract

The invention discloses a scene classification method based on interferometric synthetic aperture radar images. The scene classification method comprises the steps of 1) reading a to-be-classified interferometric synthetic aperture radar image pair including a main image and an auxiliary image, dividing the image pair into a plurality of sub-image pairs, and recording coordinates of the center positions of sub-images in the original images; 2) quickly obtaining a sub-pixel-level-precision offset and a coherence coefficient between the sub-image pairs through a fast coherence coefficient method; 3) establishing a scattered point relation graph between the offset and the center positions, and estimating an offset principal value interval based on the scattered point distribution density; 4)establishing a scattered point relation graph between the coherence coefficient and the offset on the basis of the estimated offset and coherence coefficient, and determining a coherence coefficient judgment threshold based on the scattered point distribution; and 5) based on the offset principal value interval and the coherence coefficient judgment threshold, dividing target scattering of sub-image blocks of the main image into four different types, and performing marking by using different marks to form a final classification diagram.

Description

technical field [0001] The invention relates to an image classification method, in particular to a scene classification method based on an interference synthetic aperture radar image. Background technique [0002] Interferometric Synthetic Aperture Radar (InSAR) is an effective radar imaging system, which has been widely used in the fields of digital elevation inversion and surface deformation measurement. [0003] Existing scene classification algorithms mainly use the polarization scattering information of the target, and the use of target interference information is limited. Studies have shown (document [1]: L.Farro-Famil, E.Pottier, J.-S.Lee. Unsupervised classification of natural scenes from polarimetric interferometric SARdata. Frontiers of Remote Sensing Information Processing, C.H.Chen, Ed..Singapore, Singapore : World Scientific Publishing, 105-137, 2003), in the case of vegetation or forests with various types, heights and structures in the scene, the L-band and P...

AI Technical Summary

Problems solved by technology

Studies have shown (document [1]: L.Farro-Famil, E.Pottier, J.-S.Lee. Unsupervised classification of natural scenes from polarimetric interferometric SARdata. Frontiers of Remote Sensing Information Processing, C.H.Chen, Ed..Singapore, Singapore : World Scientific Publishing, 105-137, 2003), in the case of vegetation or forests with various types, heights and structures in the scene, the L-band and P-band incoherent and coherent polarization information will be saturated in sequence. At this time, based on Classification algorithms for polarized data will not work correctly for scene classification

Images