A Physical Property-based Fusion Of Satellite Visible And Infrared Images

Geostationary satellites visible channel have a higher resolution, and texturestructure of clouds can be observed in the images. The resolution of infrared channelis Lower, but brightness temperature can be directly observed. Both images haveadvantages and disadvantages. There is some complementary to fuse the two kinds ofimages to integrate richer information.Satellite infrared images have important spectral information for meteorologicalresearch, but they have lower spatial resolution. The motivation of image fusion ofinfrared and visible (VIS) is to integrate the fine textural details of visible image andthe spectral features of infrared (IR) image to get a pseudo-high resolution IR image.For the current method of image fusion of infrared and visible has remained mostlysimple image-level, the data of both channels is synthesized as the same property twoimages. Although the fused result has excellent visual effectiveness the spectraldistortion still exists. Ignoring the actual physical property of infrared and visibleimagery in the processing method, physical property of the the fusion result cantclearly defined,thus the use of this algorithm is limited in the field of meteorology.In this paper, we propose a novel fusion method based on the thermal physicalproperty, it can take full advantage of high spatial resolution of VIS to extract itsdetails to achieve more reasonable result combined with the physical meaning of theinfrared image fusion. Firstly, the infrared image is interpolated to get sameresolution as the visible image, using multiwavelet analysis theory in image fusion toget details effectiveness and improve the image resolution after preprocess.Then theStefan-Boltzmann Law is used to make a physical correction to retain the fused imagethermal physical property，and the fused image after correction will have rensonablebrightness temperature. Finally, a series of subjective and objective evaluationmethods will be use to test the resulting images in mutual information, averagegradient, quality of image, relevance of edge information and energy variance.Qualitative and quantitative experimental results show that the proposed method notonly improves the spatial resolution, but also preserves the infrared thermal property.