Algorithmic Elimination of Parasitic Reflections in Deflectometry

Phase Measuring Deflectometry (PMD) provides far-reaching advantages for the measurement of specular freeform surfaces. Unfortunately, these benefits cannot be exploited in presence of parasitic reflections. Such reflections occur, for example, at transparent objects like progressive eyeglass lenses. A hardware solution for this problem is presented elsewhere at this conference. In this talk we present a novel, purely algorithmic solution. The basic idea is to drive the sensor with different fringe pattern frequencies (instead of, commonly, only one single frequency). Using this additional information we have sufficient data to solve an inverse problem to separate the parasitic signal originating at the rear side of the specimen from the useful signal reflected from the front surface. The still upcoming problem of local ill-conditioning is solved by combining measurements of several cameras from different directions. In order to make the measurement fast, a method to minimize the amount of required raw data was developed. We will present measuring results of unprepared blocked eyeglass lenses from the production line.