Remote Sensing Image Fusion Based On Ant Colony Edge Optimization And Minimum Hausdorff Distance Under NSST Domain

With the developing of remote sensing technology,remote sensing image data acquired by different remote sensing sensors have different spatial resolution and spectral resolution.Remote sensing image data obtained by sensors designed on the basis of optical principle are conflicting in terms of spatial detail characterization ability and spectral property retention,under the condition of the same SNR,it is necessary to sacrifice a certain spectral resolution to improve the spatial resolution of remote sensing images.Similarly,it is necessary to sacrifice spatial resolution to improve the spectral resolution of remote sensing images.In the actual research and application of remote sensing image data,we need images with both good resolutions.In order to make the remote sensing images have both better spectral characteristics and higher spatial detail characterization ability,remote sensing image fusion technology emerges at the historic moment.This paper aims at the contradiction between improving spatial resolution and preserving spectral resolution of traditional remote sensing image fusion methods and the lack of spatial detail information caused by single fusion strategy,two new algorithms are proposed:1.A remote sensing image fusion algorithm based on minimum Hausdorff distance and non-subsampled shearlet transform(NSST)is proposed.First,the first principal component of the multispectral image was obtained through principal component analysis(PCA)transformation.Then,NSST decomposition was conducted on panchromatic image and the first principal component respectively to obtain the low frequency subband and the high frequency subband with different directions and scales.The high frequency subband contains a large number of spatial details with significant features,and the traditional single fusion strategy is not suitable for all details.The fusion method using a single fusion strategy will result in the loss and mismatch of some details and spectral information in the fusion image.To solve this problem,the minimum Hausdorff distance is applied as the metric of multistrategy fusion in the high frequency subband fusion method,the corresponding fusion strategy was selected according to the similarity.The fusion strategy of spatial frequency and sparse representation is applied to the low frequency subband fusion method.Finally,the fusion image was obtained by NSST inverse transform and PCA inverse transform.By testing different satellite data sets,the results show that the proposed method is effective in preserving spectral information and improving spatial resolution.2.A remote sensing image fusion algorithm based on ant colony edge optimization and minimum Hausdorff distance under NSST domain is proposed.This algorithm is an optimization of the previous one.On the basis of multi-strategy fusion,ant colony edge optimization algorithm is adopted to preserve the complementary details of multispectral image and panchromatic image in the fusion image.Firstly,the first principal component of multispectral image was obtained by PCA transformation,and the two valued edge image of panchromatic image was obtained by ant colony edge detection of panchromatic image.According to the edge image,a new panchromatic image is obtained by multi-strategy fusion of panchromatic image and first principal component.Then,NSST decomposition was performed on the new panchromatic image and the first principal component to obtain the low frequency subband and the high frequency subband with different directional scales.The high frequency subband fusion is based on multi-strategy fusion rules,while the low frequency subband fusion method is combining spatial frequency and sparse representation.Finally,the fusion image was obtained by NSST inverse transform and PCA inverse transform.By testing different satellite data sets,the results show that the proposed method is effective in preserving spectral information and improving spatial resolution.