Swarm Intelligence for Automatic Color and Contrast Retrieval of Digital Images of Paintings

We address the following problem: given an initial high-quality reference image and a variation of it, how to compute suitable values for color map and contrast such that, when applied to this variation, we get an image very similar visually to the reference image. This problem can be formulated as an optimization problem. Unfortunately, this leads to a continuous nonlinear optimization problem too difficult to handle by classical mathematical optimization techniques. To tackle this issue, we apply a powerful swarm intelligence method called cuckoo search algorithm. The method is tested on an illustrative example of a famous painting by artist Vincent Van Gogh. The experimental results show that the method performs very well, with a similarity error rate between the reference and the reconstructed images of only 5.73%. The method can be applied to any variation of the original painting regardless of its initial color map and contrast.