SAR image classification method based on multi-scale feature learning network and bilateral filtering

Abstract

The invention discloses a SAR image classification method based on a multi-scale feature learning network and bilateral filtering, comprising the following steps: inputting high-resolution SAR image data, and normalizing image pixel values; utilizing multi-scale processing to divide the SAR image into A collection of image blocks of different scales, and realize sample expansion; select a training sample set and a test sample set from the expanded SAR image block sample set; use the training sample set to train a fully convolutional neural network to extract the features of each image block, and form Multi-scale feature representation; training softmax classifier; using the trained multi-scale feature learning network for classification; using bilateral filtering model to classify and post-process the classification result map to obtain the final classification result map. The method of the invention can not only extract rich multi-scale features of SAR images, but also expand the training sample set, thereby improving the classification accuracy of SAR images under limited data, and can be used in the classification of high-resolution SAR images.

Description

technical field [0001] The invention belongs to the field of intelligent interpretation of remote sensing images, and in particular relates to a SAR image classification method based on a multi-scale feature learning network and bilateral filtering. Background technique [0002] Synthetic Aperture Radar (SAR) is an active microwave remote sensing technology. After more than half a century of development, it has gradually developed into an important means of remote sensing and earth observation. SAR performs coherent imaging by transmitting electromagnetic pulses and receiving target echoes, and has all-day, all-weather, multi-polarization, multi-band, and high-resolution imaging capabilities. SAR image ground object classification can distinguish the type of ground objects by analyzing the acquired information such as backscattering of ground objects. It has broad application prospects in the fields of marine environment monitoring, geological resource exploration, and agric...

AI Technical Summary

Problems solved by technology

The above SAR image classification methods all need to use a large amount of data for deep network training, and the classification effect is not good when the amount of data is small

Images