The dual-band image fusion algorithm is presented based on infrared radiation characteristics. It is a more accuracy solution, in terms of the spectral, than the wavelet-based image fusion. The fusion algorithm and quality assessment is then applied to demonstrate its performance. Firstly, the gray value of thermal image is converted into corresponding radiation extiance. Secondly, the relationship between the radiation exitance and temperature in 8μm~12μm is fitted by applications of least square method. Combined with Planck blackbody radiation theory, the temperature value of different Pixel of the thermal image is obtained by calculating the radiation. Then the radiation of LWIR image scene in MWIR spectral range is derived through Planck's Formula. Thirdly, the deduced radiation, which reflects the details of the LWIR scene, is quantified and introduced into the MWIR image. Finally, the simulation of dual-band image fusion is obtained by Matlab. The results show that the objectives of image fusion not only retain the abundant spectral information of the original images, but also gain additional information by processing the dual-band data. Thus the dual-band image fusion can increase detection, recognition and identification ranges compared with the original MWIR and LWIR data.