Local directional derivative pattern for rotation invariant texture classification

Zhenhua Guo, Qin Li, Jane You, Dapeng Zhang, Wenhuang Liu

Research output: Journal article publicationJournal articleAcademic researchpeer-review

47 Citations (Scopus)

Abstract

Local binary pattern (LBP) is a simple and efficient operator to describe local image pattern. It could be regarded as a binary representation of 1st order derivative between the central and its neighbors. Based on LBP definition, in this paper, a framework of local directional derivative pattern (LDDP) is proposed which could represent high order directional derivative feature, and LBP is a special case of LDDP. Under the proposed framework, like traditional LBP, rotation invariance could be easily defined. As different order derivative information contains complementary features, better recognition accuracy could be achieved by combining different order LDDPs which is validated by two large public texture databases, Outex and CUReT.
Original languageEnglish
Pages (from-to)1893-1904
Number of pages12
JournalNeural Computing and Applications
Volume21
Issue number8
DOIs
Publication statusPublished - 1 Jan 2012

Keywords

  • Local binary pattern (LBP)
  • Local directional derivative pattern (LDDP)
  • Rotation invariance
  • Texture classification

ASJC Scopus subject areas

  • Software
  • Artificial Intelligence

Cite this