Project Proposal-3

CBIR System for Computer Art

Motivation:

In class, we saw examples of CBIR systems for finding paintings.

With a painting as a query, the system returns similar matches to the query.

However, these systems mainly focus on Impressionism or Classical paintings.

With the advent of computers, a new art form coined "Computer Art" as emerged.

Artist use code, instead of brushes and paint to make creations.

By writing code with visualization-friendly programming languages such as Processing ,

Artist create beautiful digital images and animations.

We would like to create a CBIR systems for Computer Art.

Building such a system would create a more direct approach for users to find/access Computer Art.

Implementation

A Computer Art CBIR system differs from a Classical-painting CBIR system in many aspects, which would lead to different methods for feature extraction.

• Computer art is mostly contains repetitive patterns, since it is generated from code.
• The texture of Computer art is different from the brush strokes of classical painting.
• Computer art contains geometric shapes, which would be useful for shape detection.
• etc.

With through exploration of computer art images/animations, we will discuss and think of creative ways to extract features.

Related Work:

• CBIR for Traditional Chinese Landscape http://librarystudentjournal.org/index.php/lsj/article/view/38/69
• Computer Art by John Meada http://www.maedastudio.com/ http://grahamvard.blogspot.com/2011/06/some-lovely-visuals-from-john-maeda.html