Query-by-Example Methodology

• Image retrieval by content

  • extract and compare colors
    • RGB histogram
  • texture and /or shape information
  • feature vectors using Euclidean distance
  • Probability Density Function (CANDID)
    • color content
    • localized texture and/or shape content

Previous slide

Next slide

Back to first slide

View graphic version

Borrowed from the N-gram method of comparing free-text documents

Global signature is computed for each document.

Signature is typically represented by a histogram of the number of times that each substring of length N occurs in the document

DOT PRODUCT USED TO DETERMINE THE SIMILARITY BETWEEN 2 DOCUMENTS

CANDID calculates a GLOBAL signature that represents such features as: local textures, shapes, and colors