Phase correlation

Registration of images generally means finding the geometrical transformations between the images. It is an important part of image processing since it is often necessary to compose several images together. The methods available for image registration depend on whether the images are shifted only, are similar (i.e. rotated, scaled, shifted) or are distorted in a more complicated way. In this book, Hana Druckmüllerová presents the method for image registration based on phase correlation for rotated, scaled and shifted images. The method does not require that the images to be registered have the same exposure setting, it is even not necessary that the images contain clearly visible matching structures. The method is thus extremely robust and suitable for registration tasks as challenging as the registration of coronal images taken during total solar eclipses. The complete and consistent mathematical approach to the phase correlation was developed in the whole broadness for the first time. The Fourier transform and the discrete Fourier transform and its properties necessary for the phase correlation are described.