Museum Painting Retrieval

TL;DR

This paper presents a classical computer vision-based system for retrieving museum paintings by content, evaluating various descriptors under different image perturbations to understand their robustness and effectiveness.

Contribution

It systematically compares classical feature descriptors for content-based image retrieval in museum collections under various perturbations.

Findings

MAP scores range from 0.5 to 1.0 depending on perturbations.

Color and texture descriptors perform differently under noise and color corruption.

Rotation and overlapping text impact retrieval accuracy significantly.

Abstract

To retrieve images based on their content is one of the most studied topics in the field of computer vision. Nowadays, this problem can be addressed using modern techniques such as feature extraction using machine learning, but over the years different classical methods have been developed. In this paper, we implement a query by example retrieval system for finding paintings in a museum image collection using classic computer vision techniques. Specifically, we study the performance of the color, texture, text and feature descriptors in datasets with different perturbations in the images: noise, overlapping text boxes, color corruption and rotation. We evaluate each of the cases using the Mean Average Precision (MAP) metric, and we obtain results that vary between 0.5 and 1.0 depending on the problem conditions.