A blocked statistics method based on directional derivative

Junli LI, Chengxi CHU, Gang LI, Yang LOU

Research output: Book Chapters | Papers in Conference ProceedingsConference paper (refereed)Researchpeer-review

Abstract

The basic idea of identifying the motion blurred direction using the directional derivative is that the original image be an isotropic first-order Markov random process. However, the real effect of this method is not always good. There are many reasons, of which the main is that a lot of pictures do not meet the physical premises. The shapes of objects and texture of pictures would be vulnerably influenced for identifying. In this paper, according to the image characteristics of the local variance, we extract multiple blocks and identify the motion directions of the blocks to identify the motion blurred direction. Experimental results show that our method not only improve the identification accuracy, but also reduce the amount of computation.

Original languageEnglish
Title of host publicationWeb Information Systems and Mining - International Conference, WISM 2012, Proceedings
EditorsFu Lee WANG, Jingsheng LEI, Zhiguo GONG, Xiangfeng LUO
Place of PublicationBerlin, Heidelberg
PublisherSpringer Berlin
Pages90-97
Number of pages8
Volume7529
ISBN (Electronic)9783642334696
ISBN (Print)9783642334689
DOIs
Publication statusPublished - 2012
Externally publishedYes
Event2012 International Conference on Web Information Systems and Mining, WISM 2012 - Chengdu, China
Duration: 26 Oct 201228 Oct 2012

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume7529
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference2012 International Conference on Web Information Systems and Mining, WISM 2012
Country/TerritoryChina
CityChengdu
Period26/10/1228/10/12

Keywords

  • Blocked statistics method
  • Directional derivative
  • Markov process
  • Motion blur
  • Weighted average method

Fingerprint

Dive into the research topics of 'A blocked statistics method based on directional derivative'. Together they form a unique fingerprint.

Cite this