Bag-of-Visual-Words, its Detectors and Descriptors; A Survey in Detail

Abstract

In this survey, we investigate the Bag-of-Visual-Word technique by an up-down strategy. At the beginning, we explain the general approach and functionality of the method and then we study the combination of various high level ideas and their consequent results yielded by experts and well-known authors. Subsequently, supplementary information will be provided by comparing and discussing full details of detecting and describing interest points. Moments of inertia are also studied because of their crucial role in many computer vision approaches and also their stability over some image deformation which make them a suitable tool for object recognition methods. At the end of this paper, we draw a comparison between invariant functions and covariant function through principal axis of second moments. To provide a deeper understanding, the empirical results of the comparison have been illustrated.