Synthetic aperture radar (SAR) imaging guidance has all-weather detection capability, and has been applied to aerial vehicles in the field of air combat. The research on sar/ visible image fusion technology has important theoretical and practical significance for improving the ability of target detection and precision guidance in complex combat environment. The imaging mechanism of visible light image and SAR image is different, and there are great differences in characteristics between images. Using image fusion technology to organically combine different images, complement their advantages and disadvantages, and can better interpret their scene information. Aiming at the difference feature extraction of SAR optical images, the modal classification task and ground object classification task are proposed, and the fine-tuning VGG convolution neural network model is designed, which uses less training time and achieves better classification results. For SAR optical image mapping, space-frequency consistent generation countermeasure network model framework (SFGAN) is proposed, which makes the texture details of the generated image more realistic and the contour features of roads and rivers more clear. Training from the original data, ablation experiments were designed to verify the effectiveness of the loss function. Experimental results show that the SFGAN method can learn the matching relationship from the small-scale SAR optical image pair data, so as to realize the mapping from SAR image to optical image.
Research on Image Fusion Methodof SAR and Visible Image Based on CNN
12.10.2022
1134424 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Image Fusion Processing Method Based on Infrared and Visible Light
British Library Conference Proceedings | 2019
|Elevator part fault detector based on infrared imaging and visible light image fusion
Europäisches Patentamt | 2020
|