Please use this identifier to cite or link to this item:
Title: Object Recognition by Distortion-free Graph Embedding and Random Forest
Authors: Demirci, Muhammed Fatih
Kaçka, Serdar
Keywords: Image retrieval
Object recognition
Shape context
Issue Date: 2016
Publisher: IEEE
Source: Demirci, M. F., & Kacka, S. (2016, February). Object recognition by distortion-free graph embedding and random forest. In 2016 IEEE Tenth International Conference on Semantic Computing (ICSC) (pp. 17-23). IEEE.
Abstract: Error tolerant graph matching is required not only in many realistic object recognition scenarios, but also in different domains such as document analysis and mechanical drawings. This paper presents such a technique using a distortion-free graph embedding, reformulating the problem as that of finding error-tolerant point matching in the geometric space. The embedding works by finding the distance between every node pair. In order to properly perform the embedding for object recognition in which graph nodes represent image features and graph edges show the relations between the features, we first use a machine learning algorithm, Random Forest, which obtains similar sets of features from images taken from close view points. Given a set of such features, we then perform their consistent ordering based on the local histogram around each feature. The experiments present the improved performance of the overall algorithm over the previous error tolerant matching approaches and Bag-of-visual-words model (BoVW) for object recognition.
Description: 10th IEEE International Conference on Semantic Computing (ICSC) (2016 : Laguna Hills, CA)
ISBN: 978-1-5090-0662-5
ISSN: 2325-6516
Appears in Collections:Bilgisayar Mühendisliği Bölümü / Department of Computer Engineering
Scopus İndeksli Yayınlar Koleksiyonu / Scopus Indexed Publications Collection
WoS İndeksli Yayınlar Koleksiyonu / WoS Indexed Publications Collection

Show full item record

CORE Recommender


checked on Sep 23, 2022


checked on Sep 24, 2022

Page view(s)

checked on Dec 26, 2022

Google ScholarTM



Items in GCRIS Repository are protected by copyright, with all rights reserved, unless otherwise indicated.