This paper proposed an auto-adaptive threshold method of two-dimensional (2-D) histogram based on multi-resolution analysis (MRA), decreasing the calculation complexity of 2-D histogram whereas improving the searching precision of multi-resolution threshold method. Such method originates from the extraordinary segmentation effects achieved by 2-D histogram threshold segmentation method through the spatial correlation of gray level and the flexibility as well as efficiency of the threshold searching of multi-resolution threshold segmentation method. Experiments results demonstrated that such method can obtain segmentation results similar with the exhaustive 2-D histogram method, whereas the calculation complexity decreases exponentially with the increase of resolution level.