| abstract |
The invention discloses a hyperspectral and panchromatic image fusion method based on deep learning and matrix decomposition, which mainly solves the defects of insufficient spatial information improvement and excessive spectral information loss in existing hyperspectral and panchromatic image fusion methods. The implementation steps are as follows: ①Preprocess the hyperspectral image dataset to obtain training data; ②Construct a high-frequency information convolutional network and use the training data to train it; ③Input the hyperspectral image and panchromatic image to be fused, and pass LapSRN The network super-resolves the hyperspectral image, and obtains the high-frequency details of the panchromatic image through the high-frequency information convolution network; ④ uses the panchromatic image as a guide map, and enhances the boundary information of the super-resolution hyperspectral image through guided filtering; ⑤ combines the prior The image is constructed and fused with an optimization equation, and the optimization equation is solved to obtain an output fused image. The invention reduces the loss of spectral information, improves the spatial information of images, and is suitable for the fusion of hyperspectral and panchromatic images in any scene. |