Object Recognition Using Rank Based Skeleton

Abstract

Object recognition is a subfield of computer vision, artificial intelligence, and machine learning
that seeks to recognize and identify the most prominent objects (i.e., people or things) in a
digital image. Humans perform object recognition effortlessly and instantaneously. The input
image is subjected to the formation of the rank based skeleton image as it yields the addition
data for enhancing the accuracy of image matching and it does not require manual intervention.
The rank based skeleton image is subjected to the formation of the graph that indicates the
highly significant points in the image, which is subjected to the feature extraction. The features
are fed to the modified K-Nearest Neighbor (KNN), termed as multikernel K-Nearest Neighbor
(KNN). The analysis of the methods revealed that the proposed multikernel KNN acquired the
maximal accuracy, minimal False Acceptance Rate (FAR), and minimal False Rejection Rate
(FRR), which is 0.7114, 0.2636, and 0.0266, respectively.