The addition of Superresolution Optical Fluctuation Imaging (SOFI)[1][2] to the arsenal of superresolution methods offers a simple and affordable alternative to the more sophisticated techniques (such as STED, localization microscopies, and structured illumination). Calculations of higher order cumulants lead to a larger improvement in SOFI resolution, but also introduce a large nonlinear expansion in brightness dynamic range. Balanced SOFI (bSOFI)[3] provides a way to correct for the expanded dynamic range, but at the same time it introduces artifacts in the corrected image. In this work, we introduce a novel local dynamic range compression method, where the brightness dynamic range of a high (up to 6th-) order SOFI image is locally self-calibrated by the brightness of the corresponding second order SOFI image. The method is implemented for both auto- and cross- correlation SOFI with 36 fold extra pixels, and is combined with deconvolution. Local dynamic range compression SOFI (ldrc-SOFI) is demonstrated for simulated data and for QD625 labeled alpha-tubulin in fixed 3T3 cells with a 4-fold resolution enhancement. We show that ldrc-SOFI suffers from fewer artifacts compared to bSOFI, while exhibiting faithful dynamic range compression. A large library of simulated data of filaments networks with variable filaments density, labeling density, labeling uncertainty, noise level, background level and nonspecific binding probability was generated and examined by both bSOFI and ldrc-SOFI. When the signal to noise ratio (SNR) and other sample conditions are favorable, both algorithms perform well. In fact, under very low filaments density bSOFI performs better. However, under challenging imaging conditions (high feature density, non-specific background, high noise level), ldrc-SOFI yields better performance with less artifacts.[1] PNAS, 106(52):22287, 2009.[2] Optics express, 18(18):18875, 2010.[3] Optical Nanoscopy, 1(1):1, 2012.