Shock waves in viscous flows present thick and diffuse shock layers that merge with the boundary layer. These specific characteristics make detection and analysis difficult and inaccurate. To improve shock-wave detection, a method based on Fourier self-deconvolution (FSD) is presented in this paper. Test cases consider a concrete application with a test case obtained in a supersonic rarefied flow around a cylinder. In a first step, the method was validated with numerical flow fields calculated with the DS2V code based on a Monte Carlo algorithm. Then, the FSD method was applied to experimental flows operating at Mach 2 and Mach 4 in the low-density MARHy facility. The experimental flow fields were measured using the glow discharge technique, and shock-wave detection was validated against pressure profiles measured with a Pitot tube. The proposed method makes it possible to identify three characteristic zones of the shock: the shock foot, its centre and the maximum density zone, making it possible to deduce its thickness and obtain a precise estimation of the standoff distance. In addition to being completely non-intrusive, the method appears to be parameter free, robust and easy to implement.