Real-time fusion method of panchromatic and multispectral images based on CPU and GPU co-processing

Abstract

The invention discloses a real-time fusion method of panchromatic and multispectral images based on CPU and GPU cooperative processing, which performs fast fusion processing on panchromatic multispectral data to generate fusion images with high spatial resolution and high spectral resolution. The present invention completes all the steps of panchromatic and multispectral fusion in the memory, and maps the steps with a large amount of calculation to the GPU, which can greatly improve the fusion efficiency while ensuring the fusion accuracy. The strict geometric correspondence between the chromatic and multispectral spectra, and then the possible geometric deviation between the panchromatic and multispectral spectra is limited to the maximum extent through differential correction, so that the registration effect of the panchromatic and multi-spectral can be optimized. The multi-scale SFIM fusion method and the image fusion method of panchromatic spectral decomposition can obtain the fusion image with the best effect. The present invention is suitable for fast fusion processing requirements of sensor data with large amount of data.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a method for real-time fusion of panchromatic and multispectral images based on CPU and GPU cooperative processing. Background technique [0002] From the perspective of how the sensor works, image data is divided into multispectral images and panchromatic band images. Panchromatic band image data has relatively high spatial resolution but low spectral resolution; multispectral image data has high spectral resolution but low spatial resolution due to the use of narrow band light sources for imaging. The panchromatic multi-spectral fusion method is to combine the advantages of the two, so that the fused image has both high spatial resolution and high spectral resolution, which is beneficial to later interpretation and processing. [0003] For the camera, the detection pixel, referred to as the detector, can only maintain the signal-to-noise ratio when it receives enoug...

AI Technical Summary

Problems solved by technology

[0006] In order to solve the problem of low efficiency in the process of full-color and multi-spectral fusion of large data volumes, the present invention is mainly based on CPU / GPU collaborative processing, and the entire processing flow is carried out in the memory. The CPU is responsible for process control and parallel rule division, and the GPU is responsible for parallel processing

Images