Detecting a true face includes: capturing a parallel image of a face captured by a camera having a filter polarizing parallel to a polarization direction and an orthogonal image of the face captured by a camera's filter polarizing orthogonally to the polarization direction; computing a difference image from difference between the parallel and orthogonal images; filtering the difference image in wavelet packets on 5 levels, eliminating “low” resolution levels; dividing the filtered image into sub-images; the filtered image and each sub-image undergoes: a Fourier or discreet cosine transformation; and modelling during which the frequency decrease profile is modelled by a model of the power type (a.x^b+c); extracting, from the filtered image and each sub-image, texture characteristics of face surface specularities; analyzing, for each face region, coefficient “b” and the texture characteristics extracted regarding the model corresponding to said face region; and deciding the veracity of the face from the analysis.