Correlation Filtering Theory
Vishnu Naresh Boddeti, Andres Rodriguez and B.V.K. Vijaya Kumar
Support vector machine (SVM) classiﬁers are popular in many computer vision tasks. In most of them, the SVM classiﬁer assumes that the object to be classiﬁed is centered in the query image which might not always be valid, e.g., when locating and classifying a particular class of vehicles in a large scene. In this paper we introduce a new classiﬁer called Maximum Margin Correlation Filter (MMCF), which while exhibiting the good generalization capabilities of SVM classiﬁers is also capable of localizing objects of interest, thereby avoiding the need for image centering as is usually required in SVM classiﬁers. In other words, MMCF can simultaneously localize and classify objects of interest. We test the efﬁcacy of the proposed classiﬁer on three different tasks: vehicle recognition, eye localization, and face classiﬁcation. We demonstrate that MMCF outperforms SVM classiﬁers and also well-known correlation ﬁlters.