Object Recognition with Fourier Descriptors

Abstract

This paper is concerned with simple object recognition issue when it is desired to recognize the objects due to some significant features. Fourier Descriptors (FDs) have been chosen to come up with interesting results and analysis. FDs have been used as features when identifying the objects. While making experimentations, it has been kept in mind that the introduced technique stands for the cases when the objects are seen stand alone. Some of the preprocessing steps include outline capture and noise removal of the objects. Multiple number of similarity measures have been the used presenting various experiments. The aspect of occlusion is also partially part of the work. An object database of size 100 has been chosen for testing the proposed methodology. This includes objects of variant sizes, translations, rotations, and scalings. Finally, a different solution is proposed for future research which is based on the optimization of FDs.