Face recognition by means of template matching in frequency domain-a hardware based approach

Facial images are used for automated user identification. Various algorithms are known for realizing this task. Since most of these methods are software based solutions, a hardware based design is of interest for special applications. Compared to the most common algorithms, template matching in the frequency domain is shown to be the most suitable algorithm. In the resulting solution, the mathematical operations are reduced to minimum expense. Furthermore, data compression is done by concentration on the relevant coefficients in Fourier space. Determination of the minimum Euclidian distance to one reference set of data gives the probability for a valid match.