Sparse depth network based polarization SAR (Synthetic Aperture Radar) image classification

Abstract

The invention discloses a sparse depth network based polarization SAR (Synthetic Aperture Radar) image classification method. The implementation of the sparse depth network based polarization SAR image classification method comprises step 1, inputting an image; step 2, performing filtering; step 3, extracting features; step 4, constructing and training a sparse depth network; step 5, performing prediction and classification; step 6, outputting a result. Compared with the existing method, an energy function can be minimized, a distortion function can be controlled within a certain range to enable a hidden layer to be sparse, and accordingly the calculation is greatly reduced, the generalization capability is high, and the classification accuracy is high.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to the technical field of polarimetric SAR image ground object classification. Background technique [0002] Polarization SAR is a high-resolution active microwave remote sensing imaging radar. It has the advantages of all-weather, all-time, high resolution, and side-view imaging. It can be used in many fields such as military, agriculture, navigation, and geographical surveillance. . Compared with SAR, polarimetric SAR performs full polarimetric measurement, which can obtain richer information about the target. In recent years, the classification using polarimetric SAR data has received great attention in the field of international remote sensing, and has become the main research direction of image classification. [0003] The general process of polarimetric SAR image classification includes the steps of image data acquisition, image feature extraction, training cl...

AI Technical Summary

Problems solved by technology

However, each feature extraction method has a corresponding best use occasion, and it is inevitable to lose part of the information of the original image, so it is difficult to find a particularly suitable feature

At present, the deep network has attracted extensive attention from researchers due to its powerful learning ability. It can extract more advanced features based on simple features, that is, to mine more useful information from the original information, which greatly increases the classification accuracy; but When the amount of input data is too large, the training of the network is more difficult, and the time complexity will increase a lot

Images