Edge enhancement techniques have been widely used in mapping geologic features, such as faults, contacts and dikes. The universal disadvantages of previously presented filters are that they cannot balance signals caused by geological structures located at different depths or they bring false edges or the images obtained from these filters have low resolution. We introduce a filter SF that uses the softsign function and derivatives of the horizontal gradient to solve these problems. The applicability of the filter is exemplified with three synthetic data sets and aeromagnetic data from the Olympic Peninsula, northwestern United States. Compared to other edge detector methods, the filter is able to outline the source edges more precisely and clearly, and there are no spurious lateral boundaries in the output results.