Unbalanced Polarization SAR Object Classification Method Based on Cost Sensitivity Auxiliary Learning

Abstract

The invention discloses a method for classifying unbalanced polarization SAR ground features based on cost-sensitive assisted learning. First, the polarization SAR image to be classified and the real ground object label information corresponding to the polarization SAR image are input; Take the absolute value of the polarization coherence matrix T of the SAR image data to obtain the modulus matrix |T|; then select the training sample set; build the cost-sensitive auxiliary learning model; classify the polarization SAR image after training the cost-sensitive auxiliary learning model; Finally, the visual classification result map of the whole image is output. The cost-sensitive auxiliary learning model built by the present invention not only has a simple network layer and requires less label data sets, but also can solve the problem of unbalanced ground features in the classification of polarimetric SAR ground features and achieve better classification of small-block targets , making the classification accuracy higher.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to an unbalanced polarization SAR ground object classification method based on cost-sensitive auxiliary learning. Background technique [0002] As an important research content of polarimetric SAR image understanding and interpretation, polarimetric SAR image classification has attracted more and more attention from researchers in recent years, and has been widely used in various fields, such as land cover type discrimination, ground target detection, geological exploration, etc. , identification of vegetation types, etc. According to the use of labeled samples and unlabeled samples in the classification method, the polarization SAR ground object classification methods can be divided into three types: unsupervised classification methods, supervised classification methods and semi-supervised classification methods. [0003] For polarization SAR image classificat...

AI Technical Summary

Problems solved by technology

[0007] The technical problem to be solved by the present invention is to provide a method for classification of unbalanced polarization SAR ground objects based on cost-sensitive assisted learning in view of the deficiencies in the above-mentioned prior art. Unbalanced surface object types in polarimetric SAR images are conducive to the extraction of small target information, so as to achieve the goal of improving the classification accuracy of polarimetric SAR images

Images