Feasibility study of depth classification by out-of-focus blur in an image based on CRF

Ryusuke Miyamoto, Yutaka Hirose, Shingo Kobayashi

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

First, this paper confirm whether object boudaries can be detected or not based on image blur if out-of-focus blur occurs. In this investigation, we propose a novel constructing depth classification scheme based on CRF. Experimental results using a real image show that the proposed depth classification scheme can detect object boundaries. Next, considering the results, we propose a novel object detection scheme using image blur. The proposed object detection scheme uses weak classifiers to compute the difference of a degree of blur between two rectangular regions in a weak classifier, which are based on the Haar-like features. Experimental results using synthotic images show that an object detector constructed by Adaboost learning algorithm with the weak classifiers can detect target objects based on image blur.

Original languageEnglish
Title of host publicationWMSCI 2015 - 19th World Multi-Conference on Systemics, Cybernetics and Informatics, Proceedings
EditorsNatalja Lace, Belkis Sanchez, Nagib C. Callaos, Michael Savoie, Oleg Goriachkin
PublisherInternational Institute of Informatics and Systemics, IIIS
Pages228-233
Number of pages6
ISBN (Electronic)9781941763292
Publication statusPublished - 1 Jan 2015
Event19th World Multi-Conference on Systemics, Cybernetics and Informatics, WMSCI 2015 - Orlando, United States
Duration: 12 Jul 201515 Jul 2015

Publication series

NameWMSCI 2015 - 19th World Multi-Conference on Systemics, Cybernetics and Informatics, Proceedings
Volume1

Conference

Conference19th World Multi-Conference on Systemics, Cybernetics and Informatics, WMSCI 2015
CountryUnited States
CityOrlando
Period12/07/1515/07/15

    Fingerprint

Keywords

  • CRF
  • Depth classification
  • Haar-like feature
  • Image blur
  • Object detection

Cite this

Miyamoto, R., Hirose, Y., & Kobayashi, S. (2015). Feasibility study of depth classification by out-of-focus blur in an image based on CRF. In N. Lace, B. Sanchez, N. C. Callaos, M. Savoie, & O. Goriachkin (Eds.), WMSCI 2015 - 19th World Multi-Conference on Systemics, Cybernetics and Informatics, Proceedings (pp. 228-233). (WMSCI 2015 - 19th World Multi-Conference on Systemics, Cybernetics and Informatics, Proceedings; Vol. 1). International Institute of Informatics and Systemics, IIIS.