Method and device for fusing optical image and SAR (Synthetic Aperture Radar) image

Abstract

The invention provides an optical image and SAR image fusion method and device, and the method comprises the steps: detecting the edge features of an SAR image based on phase consistency, and obtaining a first edge image of the SAR image; performing gray stretching on the first edge image to obtain a second edge image; mapping the gray levels of the corresponding points of the SAR image according to the gray levels of the points of the second edge image to obtain a strong scattering target image of the SAR image; and performing linear weighted fusion on the gray levels of the corresponding points of the strong scattering target image and the optical image to obtain a fused image.

Description

Technical field [0001] The present disclosure relates to the field of electronic information technology radar technology, in particular to a fusion method and apparatus for optical images and SAR images. Background [0002] The most widely used in remote sensing image fusion is optical imagery and synthetic aperture radar (SAR) imagery, the imaging methods of the two are different - optical imagery is passive imaging, which can obtain fine combination and texture characteristics of features, but it is not easy to find camouflage targets; SAR images are active imaging, which can observe the target all day and all day, but it is often difficult to interpret. At present, there are many methods for optical-SAR remote sensing image fusion, including airspace-weighted methods, multi-scale methods, hybrid model methods and emerging neural network methods. [0003] The traditional airspace weighting method directly weights the two images linearly, and the obtained results are easy to int...

AI Technical Summary

Problems solved by technology

[0003] The traditional spatial weighting method directly weights the two images linearly, and the obtained results are prone to mutual interference, resulting in serious distortion of the fusion result, which is difficult to interpret

Since the multi-scale method involves transformation models of different complexity, it requires a large amount of calculation in the transformation process, which is very limited in practical engineering applications.

The method based on the neural network needs to collect a large amount of sample data for model training and parameter tuning, and the initial workload is large; and the application scenarios are severely limited, and the stability of the algorithm cannot be guaranteed in actual engineering applications.

Images