A proposed method on painting authentication using Contourelet transform (PAUCT)

摘要

This paper presents a method based on contour let transform for the purpose of art authentication. The sample digital images from the different artist are converted to the gray scale images. The background patches from the gray scale images are extracted. The extracted patches are transformed using Contourelet transform and modeled using Hidden Markov Tree (HMT). The fisher distance information is calculated to measure for stylistic similarity between the brushwork of the artist. The result obtained is analyzed to determine the authenticity of the paintings. In this method we have analyzed different current existing solutions determine their limitations and propose a new hybrid method that can overcome those limitations. It is tested on the sample of twenty images of the paintings, consisting of the paintings from the Van Gog and other artist separated as the Van Gog and Non Van Gog paintings.