Fusion Evaluation Method of High Resolution Remote Sensing Image Based on Diffusion Tensor Guidance

Abstract

The present invention provides a high-resolution remote sensing image fusion evaluation method based on diffusion tensor guidance, which includes dividing the image-registered multispectral image and the panchromatic image into blocks, and using an anisotropic model to perform diffusion tensor-guided upsampling. , select the remote sensing image fusion algorithm, use GPU parallel computing acceleration, and perform correlation analysis on remote sensing image fusion evaluation indicators. In the upsampling process, the invention can adjust the intensity of the effect of the diffusion tensor model on image pixels through parameters, so that edge protection and noise removal can be balanced according to actual needs; the upsampling is not affected by the imaging quality such as the bright contrast of the original image, making the overlap The images are strictly corresponding, avoiding the false structure phenomenon of pixels near the edges. Moreover, a general model is proposed to summarize the fusion strategy, and the quality of the fused image is comprehensively evaluated through correlation analysis; the performance of the existing equipment is fully utilized, and the powerful computing advantages of the GPU are used for multi-thread processing.

Description

technical field [0001] The invention belongs to the technical field of remote sensing image fusion, in particular to a high-resolution remote sensing image upsampling fusion method based on diffusion tensor guidance. Background technique [0002] Multispectral remote sensing images play an important role in the fields of agriculture, forestry, mining, and the environment. However, due to the limitations of the radiation transfer process and sensor technology, their spatial resolution is low, and it is difficult to obtain ground object details well. On the other hand, high-resolution remote sensing satellites can obtain sub-meter-level earth observations, but their images can only provide information in the panchromatic band. Therefore, satellite platforms are often equipped with multi-spectral and panchromatic high-resolution sensors at the same time, and effectively synthesize the multi-spectral information and detailed information on the ground through image fusion to prov...

AI Technical Summary

Problems solved by technology

Existing studies have shown that while most of the current image fusion algorithms try to keep the spectral characteristics of the fused image consistent with the original multispectral image, the expressiveness of the spatial details will be reduced to a certain extent.

The reason is that most methods need to upsample the multispectral image to the same resolution as the panchromatic image in the preprocessing step. However, the commonly used upsampling methods do not take into account the geometric consistency of the multispectral image and the panchromatic image. The overlapping images are not strictly corresponding, and false structures often appear near the edge of the feature and small features

[0004] The existing remote sensing image acquisition platform is developing towards multi-satellite, multi-sensor, high spatial resolution, high spectral resolution and short return visit cycle, resulting in geometric growth of acquired remote sensing data, and traditional CPU parallel programming can no longer meet the needs of application systems for data processing. The efficiency and accuracy requirements of

Images