Polarized SAR image classification method based on DCCGAN

Abstract

The invention discloses a polarized SAR image classification method based on DCCGAN. The method comprises steps of (1) inputting an image; (2) preprocessing data; (3) normalizing a feature matrix andextracting blocks; (4) constructing a data set; (5) constructing a DCCGAN model; (6) training the DCCGAN model; (7) constructing and initializing a discriminant classification network model; (8) training the discriminant classification network model; and (9) predicting classification. The method is not required to decompose the polarized target of the polarized SAR image, can extract the featuresdirectly from a polarization scattering matrix, makes full use of the rich information of the polarized SAR image, and improves the classification precision of the polarized SAR image.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a polarimetric synthetic aperture radar based on a deep complex convolutional generative confrontation network DCCGAN (Deep Complex Convolutional Generative Adversarial Network) model in the technical field of polarimetric synthetic aperture radar image feature classification SAR (Synthetic ApertureRadar) image classification method. The invention can be used to classify ground objects in polarimetric SAR images, can effectively improve the classification accuracy of polarimetric SAR images, and can be used for object identification, tracking and positioning. Background technique [0002] Polarization SAR is a high-resolution active microwave remote sensing imaging radar, which has the advantages of all-weather, all-time, high resolution, side-view imaging, etc., and can obtain richer information on targets. The purpose of polarimetric SAR image classification is ...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that the method extracts scattering features through Pauli decomposition, does not take into account the rich information of the original polarimetric SAR complex data, and ignores some phase information of the image, resulting in insufficient utilization of image information and poor image quality. The edge of the image is detected, and the result does not reach a high classification accuracy

However, the shortcomings of this method are that DBN is not suitable for exploring local spatial correlation information in images, and randomly initializes model parameters when initializing weight parameters, resulting in slower model training and difficulty in converging to the global optimum. solution, can not get a high accuracy

Images