Unsupervised classification method based on millimeter wave complete polarization SAR images

Abstract

The invention discloses an unsupervised classification method based on millimeter wave complete polarization SAR images, and the method comprises the steps: firstly, carrying out the compression of data through a linear compression method, so that the uncalibrated data can be applied to a specific task; secondly, obtaining a category attribute of each pixel by utilizing a Wishart-H / A / alpha unsupervised classification algorithm; thirdly, employing an adaptive polarization super-pixel generation algorithm (PolASLIC) for realizing a super-pixel segmentation task so as to consider the spatial statistical characteristics of the full-polarization SAR data; and finally, fusing the spatial information obtained by the superpixels with the unsupervised pixel label information to realize a final classification task.

Description

technical field [0001] The invention relates to an unsupervised classification method based on millimeter-wave full-polarization SAR images, belonging to the field of computer vision. Background technique [0002] Synthetic Aperture Radar (SAR) is a high-resolution active imaging radar that uses sensors to obtain backscatter information from ground objects. With the continuous improvement of its performance, low-frequency SAR cannot meet the needs. Millimeter-wave SAR has the characteristics of small size, high resolution and strong electronic countermeasures, and has gradually become an important direction for the development of radar imaging. Compared with single-polarization SAR data, full-polarization SAR carries more scattering features of ground objects, and the analysis and classification of full-polarization SAR data is an important task for interpreting full-polarization SAR images. Therefore, research on millimeter-wave full-polarization SAR image classification i...

AI Technical Summary

Problems solved by technology

[0003] The classification methods of polarimetric SAR include supervised and unsupervised classification algorithms. Supervised classification methods require actual labeled samples of ground objects as training sets combined with machine learning methods to achieve supervised classification tasks, and ground truth cannot be easily obtained in many applications. , thus limiting its practical application value

Images