Abstract: In the process of using parabolic reflection to image electromagnetic interference sources, the diffraction-limited system leads to blurred and low resolution images. Interference sources can form into multi-resolution images regionally due to different frequency, so it is difficult to enhance image resolution by using existing super-resolution algorithm. This paper presents an improved Mean Shift algorithm to image adapting to multi-resolution electromagnetic interference sources. On the basis of image segmentation, we partition the multi-resolution images and use a blind deconvolution based on iterative L_R to enhance their resolution. Simulation results show that the algorithm can improve and enhance the resolution of the images formed by multi-resolution interference sources.