Scalable Object Recognition by Hallucinating Contextually Missing Features

摘要

The present invention applies a Contextual Shape Model (CSM) to an object recognition method to reconstruct a lost feature point in a context that considers the probability distribution of the size of an object and its effect on a missing feature in a context To an object recognition method. According to an aspect of the present invention, there is provided a method of reconstructing a feature point that has not been detected from an image due to disappearance in a context including a scale or an occlusion using context information, And recognizing an object from the image, the object recognition method comprising: According to an aspect of the present invention, stable and improved object recognition can be provided through reconstruction of vanished minutiae in context in a general real situation in which resolution is low or masking exists.;Object Recognition, Image Recognition, Contextual Disappearance, Contextual Shape Model, Feature Point Restoration