We present a method to implement physically an unsharp mask filter in an imaging system. The idea is based on the use of a spatially periodic variation in the detection efficiency of an area detector. Such a "structured detection" introduces harmonic peaks in the Fourier spectrum associated with the image, enabling the use of higher spatial frequencies that would otherwise be inaccessible, due to the system point spread function. The result is an effective deblurring of the image implemented in hardware, i.e., before the detection noise occurs. The method is first demonstrated via a numerical simulation and then validated using experimental neutron imaging data.