Polarized SAR Classification Method Based on NSCT and Discriminative Dictionary Learning

Abstract

The invention discloses a polarization SAR classification method based on NSCT and discriminant dictionary learning, which mainly solves the problems of low classification accuracy and slow classification speed of the existing polarization SAR image classification method. Its implementation steps are: 1. Obtain the coherence matrix of the polarimetric SAR image to be classified, and perform Lee filtering on it to obtain the denoised coherence matrix; 2. Cloude decompose the denoised coherence matrix, and decompose the 3 non-negative eigenvalues ​​and scattering angles are used as classification features; 3. Perform 3-layer non-subsampling Contourlet transformation on the classification features, and use the transformed low-frequency coefficients as the transformation domain classification features; 4. Use the transformation domain classification features, combined with the discriminant dictionary Learning model training dictionary and classifier; 5. Use the trained dictionary and classifier to classify the test samples to obtain the classification result. The invention improves classification accuracy and classification speed, and is suitable for image processing.

Description

technical field [0001] The invention belongs to the technical field of image processing, in particular to a SAR image classification method, which can be used in the field of target recognition. Background technique [0002] Synthetic Aperture Radar (SAR) is popular because of its powerful advantages of all-day, all-weather, and high-resolution. Compared with the traditional single-polarization SAR, the polarization SAR measures the target in all polarizations and can obtain richer target information, so it has attracted much attention in recent years. [0003] The images obtained by polarimetric SAR are called polarimetric SAR images. Polarimetric SAR image classification is an important research content in the process of polarimetric SAR image interpretation. Its purpose is to use the polarimetric measurement data acquired by the polarimetric SAR sensor to determine the category of each unit in the polarimetric SAR image. The results of polarimetric SAR image classificat...

AI Technical Summary

Problems solved by technology

First of all, the SRC method classifies data based on the minimum criterion of reconstruction error, but in fact there is a difference between data reconstruction and data classification. If only reconstruction error is considered without classification error, it will affect the data Classification effect; Secondly, since the SRC classifier is based on the minimum criterion of reconstruction error, each sample needs to perform multiple reconstruction error calculations, which will have an impact on the classification speed; Finally, polarimetric SAR image data due to its imaging principle The particularity makes it have strong multiplicative noise, so if it is directly classified in the original feature domain of polarimetric SAR image data, it will be disturbed by noise and affect its classification effect

Images