Universal single-time-phase and multi-time-phase SAR image speckle noise removing method

Abstract

The invention discloses a universal single-temporal and multi-temporal SAR image speckle noise removing method. The method comprises the following steps: firstly, acquiring a multi-temporal SAR image,preprocessing the multi-temporal SAR image, using a multi-temporal average image as a label image, and adding speckle noise of different degrees to generate a training sample; secondly, constructinga single-temporal SAR image speckle noise and multi-temporal SAR image speckle noise universal network model according to requirements, and fully mining any number of time phases which can be input bythe network and spatio-temporal information; determining a network training loss function, a training optimization method and hyper-parameters according to requirements; carrying out data enhancementon the training sample, wherein the data enhancement comprises normalization, cutting, overturning rotation and other operations; secondly, training a network model by using the training sample to obtain model parameters; and inputting a test sample into the network, and finally obtaining an output speckle noise removal image. The method is convenient to operate, high in calculation efficiency, low in data requirement, easy to implement, high in expandability and high in practical value.

Description

technical field [0001] The invention belongs to the technical field of remote sensing image processing, and relates to a method for removing coherent speckle noise in general single- and multi-temporal SAR images based on a deep convolutional neural network. There is a great correlation between multi-temporal SAR images, and their information is complementary Therefore, multi-temporal SAR images can be used to assist speckle removal. Background technique [0002] SAR has all-weather and all-weather earth observation capabilities, and provides a data source for long-term surface monitoring. In addition, with the successful launch of more and more SAR satellites, applications based on multi-temporal SAR images have also emerged, such as: monitoring of forests and disasters, classification of land cover, analysis of glaciers and snow, etc. In the process of SAR imaging, coherence speckle is an inevitable system noise, which seriously affects the accuracy of subsequent applicat...

AI Technical Summary

Problems solved by technology

However, most of the existing SAR coherent speckle noise removal methods based on multi-temporal phases are traditional methods, and the noise removal effect depends on manual adjustment parameters, and the utilization of spatio-temporal information is not sufficient.

Moreover, the multi-temporal method is relatively strict on the input of the number of time phases, and images of more than two time phases must be used, and the number of images also has a greater impact on the final effect.

Then the multi-temporal method cannot be used in the case of only single-temporal

Images