Panchromatic image sharpening method based on low-rank decomposition of directional multi-scale group

Abstract

The invention discloses a full-color image sharpening method based on direction multiscale group low-rank decomposition. The method comprises the steps of 1, inputting a source image; 2, obtaining LMS and dPan images; 3, calculating the spectrum correlation coefficient of a multispectral image MS and a downsampling full-color image dPan; 4, conducting non-subsample contourlet decomposition; 5, establishing a data matrix; 6, conducting matrix low-rank decomposition; 7, reconstituting a high and low frequency sparse matrix; 8, injecting the high and low frequency sparse matrix; 9, conducting non-subsample contourlet inverse transformation; 10, outputting a high-definition image. The contour structure information and details of a full-color image are extracted by means of non-subsample contourlet change and matrix low-rank decomposition, a new high and low frequency injection model is adopted, spectrum warping caused by excessive injection of the full-color image is reduced, and finally the high-definition image with better saved spectral information and more obvious edge and detail characteristics is obtained.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to remote sensing applications, a panchromatic image sharpening method based on low-rank decomposition of multi-scale groups in directions in the technical field of remote sensing image processing. The invention can be applied to the remote sensing fields such as target recognition of roads, airports, buildings, etc., and forest resources investigation. The panchromatic image sharpening is performed through the multispectral image with low resolution and high spectral rate and the panchromatic image with high resolution and low spectral rate to obtain a multispectral image with high resolution and high spectral rate. The invention can improve classification accuracy when used in remote sensing application field, and can provide clear and high-quality images in target recognition field, especially for target positioning and recognition effect of roads, buildings and the ...

AI Technical Summary

Problems solved by technology

Although this method can achieve better results, it still has the disadvantage that this method injects too much panchromatic image information, which will cause certain color distortion.

This method can obtain a better high-resolution multispectral image, but the method still has the disadvantage that the final high-resolution multispectral image loses some information of the multispectral image, which will cause certain spectral distortion. distortion

Images