Polarized SAR classification method based on shallow features and t-matrix deep learning

Abstract

The invention discloses a polarimetric SAR classification method based on shallow features and T matrix deep learning, which mainly solves the problem of low classification accuracy of the same ground object with obvious scattering information and different ground objects with similar scattering information in the prior art . The implementation steps are: 1. filter the original polarimetric SAR image; 2. extract the polarimetric shallow features of the filtered polarimetric SAR image; 3. fuse the shallow layer features with the filtered polarimetric SAR data, Construct training samples and test samples; 4. Use the convolutional neural network to learn the training samples; 5. Use the learned convolutional neural network to classify the test samples to obtain the final polarization SAR object classification results. The invention has a high accuracy rate of classification of polarized SAR target objects, and has good experimental effect on classification of large-area surface objects, and can be used for target recognition and large-scale scene object classification.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a polarimetric SAR ground object classification method, which is applicable to target recognition and large-scale ground object classification. Background technique [0002] With the rapid development of microwave remote sensing technology, high-resolution polarization synthetic aperture radar, as one of the typical representatives, will inevitably become a popular trend in the field of SAR. Although the high-resolution polarization SAR contains rich backscatter information, it has been found in practice that the complex scene information contained in real images cannot be fully expressed by only using shallow polarization features. The classification of polarimetric SAR images involves many disciplines such as physics, probability theory, pattern recognition, data mining, signal processing, etc. It is one of the important branches in the field of image processing....

AI Technical Summary

Problems solved by technology

However, the disadvantage of this method is that different feature selection is required to obtain the same good classification accuracy for different data, which obviously greatly increases the workload of researchers, and the classic SAR image features are not very important for image expression. Sufficient, this will also have a certain impact on the classification results of the method

Images