The invention discloses a rapid multi-scale super-pixel segmentation method for a synthetic aperture radar image. The method comprises the following steps: measuring the final similarity of two pixelsin combination with backscattering gray similarity, edge information similarity and homogeneity similarity; performing graph structure representation on the SAR image; constructing a minimum spanningtree for the SAR image; sAR image fast multi-scale super-pixel segmentation is achieved based on the minimum spanning tree, a super-pixel segmentation boundary is obtained based on a segmentation result, and an SAR image segmentation result is generated. According to the invention, any number of superpixels with a tree structure can be generated immediately, so that superpixel results of different levels can be segmented. According to the multi-scale super-pixel segmentation method, image boundaries can be well kept, the good image local detail keeping capacity is achieved, the segmentation result of any scale can be generated for the SAR image in real time, target information of different sizes can be easily obtained, and the execution efficiency is remarkably improved.