Home
Abstract
Process
Background
Intensity
Edges
DWT
Graininess
Color
Results
Team
 

Abstract

Given a painting from Rembrandt, Monet, or Picasso, the ArtSpy algorithm detected the artist of the painting.
Tools developed for analysis included:
  1. Intensity Analysis
          •  How dark was the painting?
  2. Edge Detection
          •  Did the artist have many edges in the painting?
  3. Discrete Wavelet Transformation Analysis
          •  Did local high frequency content characterize the artist?
  4. Graininess Analysis
          •  How grainy was the work of art?
  5. Color Analysis
          •  Did the artist paint using blocks of similar color?


Final algorithm incorporated only intensity, edge, and color analysis to determine the artist since the other two tools did not give definite trends.

Testing against 120 pictures, ArtSpy successfully detected Rembrandt with 93% accuracy (37/40), Monet with 90% accuracy(36/40), and Picasso with 78% accuracy (31/40). Overall accuracy was 87% (104/120).