Identify Occurrence of Substance Object of a Certain Classification in Fractional Videos and Pictures

Abstract

Object detection and recognition are the meta-heuristic problems in computer vision. Practically usable dynamic object recognition methods are still unavailable. A new method was proposed which improves over existing methods in every stage. In that addition features like geometric shapes, ellipsis are added. An heuristic codebook was proposed of good generalization and discriminative properties, enabling multipath interferences mechanisms on propagation of1 conditional livelihood. A new learning method also proposed which is capable of online learning.