Hartley transform based correlation filters for face recognition
Krish Narendra, Shashank Satyanarayana · 2016
This paper explores the viability of Hartley Transforms as an alternative to Fourier Transforms for Face Recognition. The paper provides a brief introduction to Hartley Transform, which is a reasonable alternate to Fourier Transform due to its similarities in the choice of basis function. Correlation filter is a pattern recognition tool that is efficient and robust. This includes extraction of features from the face images and development of various discriminant functions in the form of correlation planes. The correlation filters produce a sharp peak if an image similar to the trained image is present. The decision is based on the peak to side lobe ratio of the correlation plane. Results obtained from various classes of correlation filter indicate that Hartley Transform is a reasonable alternative to Fourier Transform.